Apparatus and method for generating pore information

Abstract

The present disclosure relates to an apparatus and a method for generating pore information, the apparatus comprising: a preprocessing unit for generating a particle mask on the basis of a cathode material cross-sectional image that includes an image of at least one particle and a plurality of pores included in the particle, and performing image restoration preprocessing on the basis of the particle mask; an image segmentation unit for extracting an individual pore mask for each pore on the basis of the cathode material cross-sectional image and the particle mask; and an information generation unit for generating pore information about the pores on the basis of the individual pore masks.

Description

Void information generation device and method

[0001] The present disclosure relates to an apparatus and method for generating void information, and more specifically, to an apparatus and method for generating void information based on a given image.

[0002] Obtaining information regarding the pores contained in a specific material can be useful for analyzing and improving products that use that material. Furthermore, the material contains various particles, and multiple pores may exist for each particle.

[0003] Considering these factors, the ratio and distribution of pores within individual particles can significantly affect the electrochemical performance of the product; therefore, the development of technology to analyze and generate pore information within the particles is required.

[0004] The present disclosure aims to provide a void information generation apparatus and method capable of generating information regarding voids based on a given image.

[0005] In one aspect, the present embodiments may provide a pore information generating device comprising: a preprocessing unit that generates a particle mask based on an image of a cross-sectional cathode material including at least one particle and an image of a plurality of pores included in the particle, and performs image restoration preprocessing based on the particle mask; an image segmentation unit that extracts individual pore masks for each pore based on the cross-sectional image of the cathode material and the particle mask; and an information generating unit that generates pore information for the pores based on the individual pore masks.

[0006] In another aspect, the present embodiments may provide a method for generating pore information, comprising: an image preprocessing step of generating a particle mask based on an image of a cathode material cross-section including at least one particle and an image of a plurality of pores included in the particle, and performing image restoration preprocessing based on the particle mask; an image segmentation step of extracting individual pore masks for each pore based on the image of the cathode material cross-section and the particle mask; and an information generation step of generating pore information for the pores based on the individual pore masks.

[0007] According to the present disclosure, a void information generating apparatus and method capable of generating information regarding voids based on a given image can be provided.

[0008] FIG. 1 is a block diagram relating to a void information generating device according to the present disclosure.

[0009] FIG. 2 is a drawing exemplarily showing a cross-sectional image according to one embodiment.

[0010] FIG. 3 is a drawing for exemplarily illustrating a configuration for determining a central particle according to one embodiment.

[0011] FIG. 4 is a drawing for exemplarily illustrating a configuration for determining center coordinates according to one embodiment.

[0012] FIG. 5 is a diagram illustrating, by way of example, a configuration for performing restoration preprocessing according to one embodiment.

[0013] FIG. 6 is a drawing for exemplarily illustrating a configuration for performing smoothing processing according to one embodiment.

[0014] FIG. 7 is a diagram illustrating, by way of example, a configuration for performing critical processing according to one embodiment.

[0015] FIG. 8 is a diagram illustrating, by way of example, a configuration for performing pixel connectivity analysis according to one embodiment.

[0016] FIG. 9 is a drawing for exemplarily illustrating a configuration for generating void information according to one embodiment.

[0017] FIG. 10 is a graph illustrating, for example, a configuration for performing void information visualization according to one embodiment.

[0018] FIG. 11 is a flowchart relating to a method for generating void information according to one embodiment.

[0019] FIG. 12 is a flowchart relating to an image preprocessing step according to the present disclosure.

[0020] FIG. 13 is a flowchart relating to an image segmentation step according to one embodiment.

[0021] FIG. 14 is a flowchart relating to an information generation step according to one embodiment.

[0022] Hereinafter, some embodiments of the present disclosure will be described in detail with reference to the exemplary drawings. In assigning reference numerals to the components of each drawing, the same components may have the same reference numeral as much as possible, even if they are shown in different drawings. Furthermore, in describing the embodiments, if it is determined that a detailed description of related known components or functions may obscure the essence of the technical concept, such detailed description may be omitted. Where terms such as "comprising," "having," or "consisting of" are used in this specification, other parts may be added unless "only" is used. Where a component is expressed in the singular, it may include a plural unless otherwise specified.

[0023] Additionally, terms such as first, second, A, B, (a), (b), etc., may be used to describe the components of the present disclosure. These terms are used merely to distinguish the components from other components, and the nature, order, sequence, or number of the components are not limited by such terms.

[0024] In describing the positional relationship of components, where it is stated that two or more components are "connected," "combined," or "joined," it should be understood that while the two or more components may be directly "connected," "combined," or "joined," they may also be "connected," "combined," or "joined" with other components "intervened." Here, the other components may be included in one or more of the two or more components that are "connected," "combined," or "joined" with one another.

[0025] In describing the temporal flow relationship regarding components, methods of operation, or methods of production, for example, when the temporal or sequential relationship is described using "after," "following," "next," or "before," it may include cases where the relationship is not continuous unless "immediately" or "directly" is used.

[0026] Meanwhile, where numerical values ​​or corresponding information regarding a component (e.g., levels, etc.) are mentioned, even without separate explicit notation, the numerical values ​​or corresponding information may be interpreted as including a range of error that may occur due to various factors (e.g., process factors, internal or external shocks, noise, etc.).

[0027]

[0028] FIG. 1 is a block diagram relating to a void information generating device according to the present disclosure.

[0029] Referring to FIG. 1, the void information generating device (100) according to the present disclosure may include a preprocessing unit (110), an image segmentation unit (120), and an information generating unit (130). The preprocessing unit (110), the image segmentation unit (120), and the information generating unit (130) may be connected to each other.

[0030] For example, a void information generating device (100) may include a preprocessing unit that generates a particle mask based on an image of a cathode material cross-section including at least one particle and an image of a plurality of voids included in the particle, and performs image restoration preprocessing based on the particle mask; an image segmentation unit that extracts individual void masks for each void based on the image of the cathode material cross-section and the particle mask; and an information generating unit that generates void information for the voids based on the individual void masks.

[0031] For example, the preprocessing unit (110) can generate a particle mask based on a cross-sectional image of the cathode material. For example, if there is only one particle included in the cross-sectional image of the cathode material, a particle mask can be generated for that particle.

[0032] As another example, the preprocessing unit (110) can determine one particle using a preset criterion when there are two or more particles included in the cross-sectional image of the cathode material, and generate a particle mask for the determined particle.

[0033] For example, the preprocessing unit (110) can determine a central particle located at the center based on the entire cross-sectional image of the cathode material and generate a particle mask for the central particle. In this case, the preprocessing unit (110) can determine the central coordinates of the cross-sectional image of the cathode material and determine the particle in the area containing the central coordinates as the central particle.

[0034] For example, a cross-sectional image of the cathode material may include an image generated using a scanning electron microscope (hereinafter referred to as 'SEM'). For example, the cross-sectional image of the cathode material may include an image obtained by imaging the sample surface using an SEM after undergoing a process of segmenting or slicing a desired region while scanning the surface of the cathode active material with a Focused Ion Beam (FIB).

[0035] For example, a cross-sectional image of an anode material may include areas where a curtaining effect and contamination have occurred. For instance, a curtaining effect may occur in the cross-sectional image of an anode material as brightness changes at each depth during the process of imaging the surface of the anode material sample.

[0036] In some cases, the curtain effect may occur due to differences in roughness and density that arise during the process of irradiating the sample surface with a straight ion beam in FIB processing. Additionally, cross-sectional images of the cathode material may be contaminated with artifacts caused by contaminants surrounding the sample.

[0037] To resolve issues such as curtain effects or contamination in images, methods such as using liquid nitrogen to slow down the contamination reaction or measuring cross-sectional images as quickly as possible immediately after splitting the sample surface exist; however, since optimal conditions vary depending on the material, this can be difficult to resolve in some cases.

[0038] Considering these points, apart from methods to resolve curtain effects or contamination issues during the process of generating cross-sections and capturing images of the cathode material, it is necessary to accurately generate pore information while eliminating the effects caused by curtain effects or contamination when they are present in the captured cross-section image of the cathode material.

[0039] For example, the preprocessing unit (110) can set guide coordinates based on a cross-sectional image of the cathode material, determine particle center coordinates based on the guide coordinates using a preset SAM model, and generate a particle mask based on the particle center coordinates.

[0040] For example, the preprocessing unit (110) can determine the particle center coordinates based on preset guide coordinates and generate a particle mask based on the cathode material cross-sectional image and the particle center coordinates. For example, the particle mask can be generated by extracting a portion corresponding to the area containing the central particle from the cathode material cross-sectional image.

[0041] For example, the preprocessing unit (110) can generate a particle mask based on particle center coordinates generated for a cross-sectional image of an anode material. In some cases, the preprocessing unit (110) can determine the particle maximum radius based on the particle mask and the particle center coordinates. For example, the distance between the particle center coordinates and the pixel furthest from the particle center coordinates among the pixels included in the particle mask can be determined as the particle maximum radius.

[0042] For example, the particle mask may be a circular area with the maximum particle radius as its radius. As another example, the particle mask may be an area that replicates the particle shape shown in the cross-sectional image of the cathode material; in this case, additional preprocessing may be performed to replicate the particle shape. In this case, the mask corresponding to the circular area can be set as the first particle mask, and the mask corresponding to the area replicating the particle shape can be set as the second particle mask.

[0043] For example, the preprocessing unit (110) can use an artificial intelligence model to recognize and extract a specific object from an image or generate a segmented image. For example, the preprocessing unit (110) can use an artificial intelligence model to recognize and extract a specific particle from a cross-sectional image of an anode material or generate a segmented image. In this case, the extracted or segmented image can be generated in the form of a mask, and the extracted or segmented image in the form of a mask can be referred to as a segmentation mask.

[0044] For example, the preprocessing unit (110) can generate a particle mask for a specific particle in a cross-sectional image of the cathode material using an artificial intelligence model. In this case, the preprocessing unit (110) can generate a particle mask for a single particle appearing in the cross-sectional image of the cathode material. Alternatively, a particle mask can be generated for each particle appearing in the cross-sectional image of the cathode material.

[0045] For example, the artificial intelligence model used in the preprocessing unit (110) may include a foundation model capable of zero-shot image segmentation. For example, it may include a SAM (Segment Anything Model, hereinafter referred to as the 'SAM model').

[0046] For example, the preprocessing unit (110) can segment an image in a zero-shot manner using a SAM model. For example, the SAM model can recognize objects within an image and perform image segmentation even when data completely different from the previously trained data is input.

[0047] For example, the preprocessing unit (110) can generate a segmentation mask for a specific object based on the SAM model and preset guide information.

[0048] For example, the preprocessing unit (110) can set guide coordinates based on a cross-sectional image of the cathode material. For example, the coordinates of the center point of the cross-sectional image of the cathode material can be set as guide coordinates.

[0049] For example, the preprocessing unit (110) can determine the particle center coordinates based on the cross-sectional image of the cathode material and guide coordinates. In some cases, the preprocessing unit (110) can determine the particle center coordinates based on the cross-sectional image of the cathode material and guide coordinates, but using a SAM model.

[0050] For example, the particle center coordinates can be determined using the following [Equation 1].

[0051]

[0052] Here, cx and cy can represent guide coordinates, and dx and dy can represent particle center coordinates. According to [Equation 1], particle center coordinates dx and dy can be determined using the SAM model within a horizontal and vertical range of 40 pixels each, based on the guide coordinates cx and cy. In this case, argmax may include an operation that returns the index of the element with the largest value in the array or matrix represented by SAM(cx+dx, cx+dy).

[0053] In some cases, the pixel range for determining the particle center coordinates based on the guide coordinates may be changed to a range different from 40 pixels. For example, if the size of the central particle is relatively large or the shape of the central particle deviates significantly from a circle, the pixel range in [Equation 1] may be changed to a larger size and used to determine the particle center coordinates.

[0054] For example, the preprocessing unit (110) can generate a particle mask based on particle center coordinates. In this case, the preprocessing unit (110) can generate a particle mask based on particle center coordinates, but using a SAM model.

[0055] For example, the preprocessing unit (110) can determine a reliability score when generating a particle mask using a SAM model. For example, the reliability score may include a number between 0 and 1 representing the result of determining how accurately particle objects appearing in the cross-sectional image of the cathode material were extracted to generate the particle mask when generating a particle mask based on a cross-sectional image of the cathode material.

[0056] For example, the preprocessing unit (110) can restore an image by removing parts from the original image that are not necessary for image analysis and generate a restored image. For example, the preprocessing unit (110) can generate a restored image by performing image restoration preprocessing to remove curtain effects and contamination from the original image.

[0057] For example, the preprocessing unit (110) can generate a restored image through image restoration preprocessing to remove curtain effects and contamination, etc. from a cross-sectional image of an anode material. As another example, the preprocessing unit (110) can generate a restored image by performing image restoration preprocessing on a particle mask.

[0058] For example, the preprocessing unit (110) can perform image restoration preprocessing based on a preset VSNR algorithm and energy function for the particle mask.

[0059] For example, the Visual Signal-to-Noise Ratio (VSNR) algorithm may involve quantitatively evaluating image quality. Specifically, the VSNR algorithm may be performed by defining an energy function representing image quality and finding an image that minimizes the energy function value.

[0060] For example, a VSNR algorithm may include setting a particle mask as the original image and generating a reconstructed image based on an energy function and the original image.

[0061] For example, the energy function may include a data term and a smoothing term. For example, the energy function may include a data term set based on the difference between the original image and the reconstructed image, and a smoothing term set based on the gradient of the reconstructed image. In this case, the data term may include a measurement formula for the difference between the original image and the reconstructed image, and the smoothing term may include a control formula for adjusting the smoothness of the image.

[0062] In some cases, the energy function may further include a regularization term that can be used to control model complexity and prevent model overfitting using a pre-set regularization function. For example, the regularization function may be generated based on Total Variation (TV) based on the sum of variations, Smooth Total Variation (STV) based on the sum of variations after smoothing, or Huber Regularization based on the sum of variations determined by considering the magnitude of the gradient.

[0063] For example, the energy function can be exemplarily represented using the following [Equation 2].

[0064]

[0065] Here, Ω is the domain of the image, The part (dotted square) indicates that integration is performed over the entire Ω region in setting the integration range, where f represents the original image, u represents the reconstructed image, ∇u represents the gradient of the image, Ø represents a function defining the regularization term, and λ represents a parameter that balances the two terms within [Equation 2].

[0066] and The term can indicate how close the restored image u is to the original image f.

[0067] also, The term can represent the degree of smoothness of image u based on the gradient.

[0068] For example, Ø(∇u) can be used to remove noise while preserving edges appearing in an image. In this case, functions such as TV (Total Variation), STV (Smooth Total Variation), and Huber Regularization can be used for Ø(∇u).

[0069] For example, the preprocessing unit (110) can perform image restoration preprocessing in a direction in which the applied value of the energy function decreases using a preset Euler-Lagrange equation and stochastic gradient descent (SGD).

[0070] As a specific example, the preprocessing unit (110) can perform preprocessing using the VSNR algorithm by deriving the Euler-Lagrange equation using a variational method and performing it through an optimization method such as stochastic gradient descent (SGD).

[0071] For example, the Euler-Lagrange equation can provide mathematical conditions for minimizing the energy function. For instance, the Euler-Lagrange equation is as follows:

[0072] It can be exemplarily represented using [Mathematical Formula 3].

[0073]

[0074] Here, f can represent the original image, u the restored image, and div the divergence operator.

[0075] For example, the preprocessing unit (110) can determine the gradient of the energy function or update related parameters using gradient descent. And in some cases, the above operations can be performed using stochastic gradient descent.

[0076] For example, stochastic gradient descent may include generating a mini-batch using only a subset of the total data, and using the generated mini-batch to determine the gradient of the energy function or update related parameters.

[0077] The image segmentation unit (120) may be connected to at least one of the preprocessing unit (110) and the information generation unit (130). In this case, new information may be generated based on information received from the preprocessing unit (110), or the received information may be regenerated using other information or preset values. The generated information may then be provided to the information generation unit (130).

[0078] For example, the image segmentation unit (120) can perform Gaussian smoothing on the particle mask.

[0079] For example, Gaussian smoothing may include determining a Gaussian kernel and a sigma value based on the radius of the central particle. By using Gaussian smoothing, noise remaining in the particle mask even after preprocessing is performed can be removed.

[0080] For example, the image segmentation unit (120) can perform Gaussian smoothing by determining the Gaussian kernel and sigma value to be relatively large when the radius of the central particle is relatively large.

[0081] As another example, the image segmentation unit (120) may determine that there is a great need to preserve edges between pores when there are many individual pores included in the particle mask or many overlapping pores, and based on this determination, may perform Gaussian smoothing by determining a relatively small Gaussian kernel and sigma value.

[0082] For example, the image segmentation unit (120) can perform threshold processing to binarize the particle mask based on a preset gap threshold.

[0083] In some cases, the image segmentation unit (120) can perform local threshold processing. For example, threshold processing can be performed by setting different void threshold values ​​for each region of the particle mask.

[0084] For example, the image segmentation unit (120) can perform local thresholding by setting a different gap threshold for each region when the brightness distribution in each part of the particle mask is not uniform or when the difference in brightness between the background and the object is relatively large.

[0085] In this case, the image segmentation unit (120) can determine a void threshold value for each region based on the result of dividing the particle mask into arbitrary regions and extracting average brightness information, brightness change information, etc. for each region.

[0086] In this way, when threshold processing is performed on the particle mask, the recognition of images of individual pores included in the particle mask and the extraction of individual pore masks can be performed more accurately and quickly. Furthermore, by considering the distribution of individual pores, quantitative information regarding the distribution of individual attacks can be generated more easily.

[0087] For example, the image segmentation unit (120) can perform pixel connectivity analysis to recognize pixels connected to each other within a particle mask and group them by object, and determine whether to extract each individual void mask based on the analysis results.

[0088] For example, the image segmentation unit (120) can perform pixel connectivity analysis on the particle mask. For example, pixel connectivity analysis may include recognizing pixels included in the image as a single object based on pre-set connectivity criteria. In some cases, pixel connectivity analysis may be performed on an image that has undergone binarization processing.

[0089] For example, binarization processing may include global thresholding based on a single threshold for the entire image, local thresholding based on setting a local threshold for each region of the image, Bernsen's Method using a threshold determined based on the mean and standard deviation of each region of the image, and Niblack's Method using a threshold determined based on the mean and a fixed constant of each region of the image.

[0090] For example, when performing pixel connectivity analysis on a cross-sectional image of an anode material containing multiple void images, the size or shape of each void may differ; therefore, the accuracy of recognizing and segmenting each void can be improved through pixel connectivity analysis.

[0091] For example, the image segmentation unit (120) can generate individual masks by specifically extracting some of the regions included in the particle mask. In some cases, the image segmentation unit (120) can generate individual masks by creating a new image identical to the image corresponding to a specific partial region of the particle mask while preserving the original image of the particle mask.

[0092] For example, the image segmentation unit (120) can specify a partial region for each void included in the particle mask, and extract the specified partial region for each void to generate individual void masks. The configuration for generating these individual void masks will be explained in more detail below in FIG. 8.

[0093] The information generation unit (130) can generate pore information for the pores based on individual pore masks. For example, the information generation unit (130) can generate individual pore location information and individual pore size information for each individual pore mask based on individual pore masks. For example, the information generation unit (130) can generate pore ratio information and pore distribution information based on particle masks and individual pore masks.

[0094] For example, pore distribution information may include pore number distribution information and pore size distribution information.

[0095] For example, the information generation unit (130) can generate individual void information based on individual void masks. In this case, the individual void information may include individual void location information and individual void size information.

[0096] For example, individual void location information may include information regarding the location of each void existing within the particle mask. For example, the individual void location information may include information expressing the location of the void relative to the center of the particle in the particle mask in the form of (angle, radius). For another example, the individual void location information may include information expressed in the form of (X-coordinate, Y-coordinate) within the particle mask.

[0097] For example, individual pore size information may include information regarding the size of each pore existing within the particle mask. For example, individual pore size information may include information in a format expressed as a relative ratio based on the size of the particle mask. For another example, individual pore size information may include information in a format that estimates the absolute size of each pore based on the actual size of the particle represented by the particle mask.

[0098] For example, the information generation unit (130) can generate total pore information based on individual pore information. In this case, the total pore information may include pore ratio information and pore distribution information. Additionally, depending on the case, the information generation unit (130) can generate total pore information by further considering particle masks and individual pore masks in addition to individual pore information.

[0099] For example, void ratio information may include information regarding the ratio of each void classified according to preset criteria.

[0100] For example, the pore ratio information may include information representing the total area occupied by individual pores as a ratio to the total area of ​​the particle mask.

[0101] As another example, the void ratio information may include information representing the number of voids at each distance as a ratio to the total number of individual voids, calculated by counting the distance to each void relative to the central particle of the particle mask.

[0102] As another example, void ratio information may include information representing the ratio of the number of voids included in each size set to the total number of individual voids, after classifying void sets by the size of each void.

[0103] For example, the pore distribution information may include information regarding the distribution of each pore classified according to preset criteria. In some cases, the pore distribution information may include at least one of pore number distribution information and pore size distribution information.

[0104] For example, the void number distribution information may include information regarding the number of individual voids distributed according to the distance to each void relative to the central particle of the particle mask.

[0105] As another example, the pore count distribution information may include information regarding the number of individual pores distributed in each of the at least two sub-regions included in the particle mask.

[0106] For example, the void size distribution information may include information indicating the number of voids included in each size set after classifying void sets by the size of each void.

[0107] For example, the information generation unit (130) can generate visualization information based on pore distribution information. In this case, the visualization information may include information that visualizes the distribution of pores included in the particle mask. For example, it may include graph information regarding the pore distribution, scatter plot information showing the pore distribution scattered in the form of points, histogram information regarding the pore distribution, etc.

[0108] For example, the information generation unit (130) can generate histogram information based on at least one of the void number distribution information and void size distribution information.

[0109] For example, the information generation unit (130) can generate histogram information based on the distribution information of the number of voids. For example, the information generation unit (130) can generate histogram information regarding a histogram in which the distance to each void relative to the center particle of the particle mask is the X-axis and the number of individual voids for each distance is the Y-axis.

[0110] For example, the information generation unit (130) can generate histogram information based on void size distribution information. For example, the information generation unit (130) can classify void sets by the size of each void, and then generate histogram information regarding a histogram in which each size set is on the X-axis and the number of individual voids for each size set is on the Y-axis.

[0111] For example, the information generation unit (130) can generate void information for individual void masks. In this case, the void information may include quantitative information regarding the ratio, size, and distribution of the voids.

[0112] For example, the individual pore mask may be image data obtained through a process of generating a particle mask by extracting central particles from a cross-sectional image of the cathode material, a process of removing curtain effects and contamination using a VSNR algorithm or the like on the particle mask, and a process of dividing the image into images for individual pores through Gaussian smoothing and local thresholding.

[0113] Since the image data described above clearly distinguishes individual pores, information regarding the ratio and size of the pores can be analyzed and quantified. Furthermore, depending on the case, the process of generating this pore information can be automated. That is, upon inputting a given image, information on the size and ratio of pores, size information by radius within the particle mask, and pore distribution information can be generated.

[0114] As another example, void distribution information may be provided in the form of a histogram as well as text. For instance, visualized information can be provided by histogramming the results of counting the number of voids at each radius within the maximum radius range of the particle mask, relative to the center of the particle. In this case, the X-axis of the histogram can represent the radius of a circle corresponding to the distance from the particle center to the void, and the Y-axis can represent the result of counting the number of voids for each radius.

[0115] For example, the information generating unit (130) can obtain information related to the shape of the positive active material from a cross-sectional image of the positive active material. In this case, the obtained information may include information regarding numerical values ​​such as the ratio of voids within the particles, average size, and distribution of voids.

[0116] Through this, contamination and curtain effects can be removed from cross-sectional images of the cathode active material, allowing for more accurate segmentation of pores. In some cases, the pore segmentation images and pore information obtained through this process can be used to provide more reliable information for secondary battery design and analysis.

[0117] FIGS. 2 to 5 are drawings for explaining a configuration for performing image preprocessing using a void information generating device according to one embodiment.

[0118] FIG. 2 is a drawing exemplarily showing a cross-sectional image according to one embodiment.

[0119] Referring to FIG. 2, a void information generating device according to one embodiment can receive a cross-sectional image of an anode material and can generate void information based on the cross-sectional image of the anode material.

[0120] For example, the cross-sectional image of the cathode material may include a cross-sectional image of one particle. And in some cases, the cross-sectional image of the cathode material may include cross-sectional images of two or more particles.

[0121] For example, the cross-sectional image of the cathode material may include cross-sectional images of six particles, particles 01 to 06. For example, the cross-sectional image of the cathode material may include a cross-sectional image of particle 01 (210), a cross-sectional image of particle 02 (220), a cross-sectional image of particle 03 (230), a cross-sectional image of particle 04 (240), a cross-sectional image of particle 05 (250), and a cross-sectional image of particle 06 (260).

[0122] In this case, the cross-sectional images of each particle may be images containing voids, contamination, curtain effects, noise, etc., respectively. Additionally, depending on the case, void information may be generated for each particle individually, or void information may be generated only for specific particles.

[0123] For example, a void information generating device according to one embodiment can generate void information for a cross-sectional image of a particle located at the center of a cross-sectional image of an anode material. In FIGS. 3 to 10 below, an embodiment is described in which a central particle is determined in a cross-sectional image of an anode material and void information is generated for a cross-sectional image portion of the central particle.

[0124] FIG. 3 is a drawing for exemplarily illustrating a configuration for determining a central particle according to one embodiment.

[0125] Referring to FIG. 3, a void information generating device according to one embodiment can determine a central particle of a cross-sectional image of an anode material. Then, a void information generating process disclosed in FIG. 4 to FIG. 10 can be performed on a central particle cross-sectional image (310) of the determined central particle.

[0126] For example, a void information generating device may display a cross-sectional image of an anode material in the form of a grid, determine a central grid (320) among the grids, and determine a particle containing the central grid as a central particle.

[0127] As another example, the void information generating device can generate a central horizontal axis (330) located at the midpoint of the cross-sectional image of the anode material among the horizontal axes and a central vertical axis (340) located at the midpoint of the cross-sectional image of the anode material among the vertical axes, and then determine a particle containing the intersection point of the central horizontal axis (330) and the central vertical axis (340) as a central particle.

[0128] FIG. 4 is a drawing for exemplarily illustrating a configuration for determining center coordinates according to one embodiment.

[0129] Referring to FIG. 4, a void information generating device according to one embodiment can set guide coordinates (412) based on a cross-sectional image (410) of an anode material, determine particle center coordinates (422) based on a preset SAM model and guide coordinates (412), and generate a particle mask (420) based on the particle center coordinates (422).

[0130] For example, the void information generating device can set guide coordinates (412) based on the cross-sectional image (410) of the anode material. For example, the coordinates located at the exact center of the cross-sectional image (410) of the anode material can be set as guide coordinates (412).

[0131] For example, the void information generating device can determine the particle center coordinates (422) of the central particle using a SAM model based on the cathode material cross-sectional image (410) and guide coordinates (412). For example, the particle center coordinates (422) can be determined using the aforementioned [Equation 1].

[0132] For example, the void information generating device can generate a particle mask (420) based on the particle center coordinates (422). In some cases, the void information generating device can generate the particle mask (420) by further using the particle maximum radius (424).

[0133] For example, the void information generating device can determine the distance to the point furthest from the center particle based on the particle center coordinates (422) as the particle maximum radius (424). Then, a particle mask (420) can be generated by extracting a circular area with the radius as the particle maximum radius (424) centered on the particle center coordinates (422) from the cathode material cross-sectional image (410).

[0134] As another example, the pore information generating device may generate a first particle mask and a second particle mask. For example, a mask based on a circular area, such as the particle mask (420) shown in FIG. 5, may be generated as the first particle mask, and a mask based on the actual area of ​​a central particle contained within the first particle mask may be generated as the second particle mask. Then, the pore information generating device may generate pore information using at least one of the first particle mask and the second particle mask.

[0135] FIG. 5 is a diagram illustrating, by way of example, a configuration for performing restoration preprocessing according to one embodiment.

[0136] A pore information generating device according to one embodiment can perform image restoration preprocessing on a particle mask. In this case, the pore information generating device can perform image restoration preprocessing based on a preset VSNR algorithm and an energy function.

[0137] For example, the pore information generating device can perform image restoration preprocessing on the particle mask (610) before preprocessing and can remove all or part of the contamination and curtain effect portions. And through this image restoration preprocessing, a particle mask (530) after preprocessing can be generated.

[0138] The following describes the process of performing image restoration preprocessing on an enlarged partial image of a specific part of a particle mask.

[0139] Referring to FIG. 5, when performing image restoration preprocessing using a void information generating device according to one embodiment, a preprocessed partial image (532) can be generated based on a preprocessed partial image (512).

[0140] And, as shown in FIG. 5, the pre-processing partial enlarged image (520) is an enlarged version of the pre-processing partial image (512), and the pre-processing partial enlarged image (540) is an enlarged version of the pre-processing partial image (532).

[0141] For example, the magnified image (520) of the pre-processing portion may include not only voids but also contaminants and curtain effect portions. For example, the magnified image (520) of the pre-processing portion may include contaminants and curtain effect portions (522) of the pre-processing portion.

[0142] For example, image restoration preprocessing may be performed on a partial enlarged image (520) before preprocessing to generate a partial enlarged image (540) after preprocessing. In this regard, when comparing the images before and after the image restoration preprocessing, changes may occur such that the images of the contamination and curtain effect parts (542) after preprocessing become relatively fainter or edges are partially removed compared to the contamination and curtain effect parts (522) before preprocessing. Through this, it is possible to prevent all or part of the contamination or curtain effect parts that do not correspond to voids within the particle mask from being mistaken for voids.

[0143] FIGS. 6 to 8 are drawings for explaining a configuration for performing image segmentation using a void information generating device according to one embodiment.

[0144] FIG. 6 is a drawing for exemplarily illustrating a configuration for performing smoothing processing according to one embodiment.

[0145] A void information generating device according to one embodiment can perform Gaussian smoothing on a particle mask. For example, Gaussian smoothing may include determining a Gaussian kernel and a sigma value based on the radius size of a central particle. By using Gaussian smoothing, noise remaining in the particle mask even after preprocessing is performed can be removed.

[0146] The following describes the process of performing Gaussian smoothing on an enlarged partial image of a specific part of a particle mask.

[0147] Referring to FIG. 6, when performing Gaussian smoothing using a void information generating device according to one embodiment, a partial image after smoothing (632) can be generated based on a partial image before smoothing (612).

[0148] And, as shown in FIG. 6, the enlarged image of the part before smoothing (620) is an enlarged version of the part before smoothing (632), and the enlarged image of the part after smoothing (640) is an enlarged version of the part after smoothing (632).

[0149] For example, the enlarged image (620) before smoothing may contain not only voids but also noise remaining after preprocessing. For example, the enlarged image (620) before smoothing may contain noise portions (622) before smoothing.

[0150] For example, Gaussian smoothing may be performed on a partial enlarged image (620) before smoothing to generate a partial enlarged image (640) after preprocessing. In this regard, when comparing the images before and after Gaussian smoothing, a change may occur in which the images of the noise parts (642) after smoothing are all or partly removed compared to the noise parts (622) before smoothing. Through this, it is possible to prevent all or part of the noise parts that do not correspond to voids within the particle mask from being mistaken for voids.

[0151] For example, Gaussian smoothing may involve determining the Gaussian kernel and sigma values ​​based on the radius of the central particle. For instance, if the radius of the central particle is relatively large, Gaussian smoothing may be performed by determining the Gaussian kernel and sigma values ​​to be relatively large. As another example, if there are many individual voids or many overlapping voids included in the particle mask, Gaussian smoothing may be performed by determining the Gaussian kernel and sigma values ​​to be relatively small.

[0152] FIG. 7 is a diagram illustrating, by way of example, a configuration for performing critical processing according to one embodiment.

[0153] Referring to FIG. 7, a void information generating device according to one embodiment may perform threshold processing to binarize a particle mask (710) based on a preset void threshold value. In this case, the particle mask (710) may have undergone image restoration preprocessing and Gaussian smoothing.

[0154] In some cases, the pore information generating device according to one embodiment may perform local threshold processing. For example, threshold processing may be performed by setting different pore threshold values ​​for each region of the particle mask.

[0155] For example, a void information generating device according to one embodiment can perform local threshold processing by setting a different void threshold for each region when the brightness distribution in each part of the particle mask is not uniform or when the difference in brightness between the background and the object is relatively large.

[0156] In this case, the void information generating device according to one embodiment can determine a void threshold value for each region based on the result of extracting average brightness information, brightness change information, etc. for each region by dividing the particle mask into arbitrary regions.

[0157] For example, the particle mask (710) can be changed into either a white area or a black area after threshold processing is performed. For example, the background area without particular features within the particle mask (710) can be changed to a white area, and the voids (712) can be changed to a black area. This makes it easier to perform image segmentation for individual voids in the particle mask (710).

[0158] FIG. 8 is a diagram illustrating, by way of example, a configuration for performing pixel connectivity analysis according to one embodiment.

[0159] A void information generating device according to one embodiment can perform pixel connectivity analysis to recognize interconnected pixels within a particle mask and group them by object. Then, based on the results of such analysis, it can determine whether to extract each individual void mask.

[0160] Referring to FIG. 8, through pixel connectivity analysis, each void (812, 814, 816) included in the particle mask (810) can be recognized as a single object and can be grouped by pixels recognized as being included in the same object. These pixel groups can be distinguished and displayed by different colors or brightness.

[0161] Through this, individual pore masks can be generated by more clearly distinguishing images for each pore included in the particle mask (810). For example, the pore information generating device can generate individual pore masks (822, 824, 826) based on images for each pore corresponding to each pore (812, 814, 816) included in the particle mask (810).

[0162] For example, a void information generating device can generate individual masks by specifically extracting some regions among the regions included in a particle mask. In some cases, the void information generating device can generate individual masks by creating a new image identical to the image corresponding to a specific partial region of the particle mask while preserving the original image of the particle mask.

[0163] For example, individual pore masks can be created by specifying a partial region for each pore within the particle mask (810) and extracting the specific partial region for each pore.

[0164] To explain with a more specific example, when individual pore masks are generated for pore_001 (812), pore_002 (814), and pore_003 (816), which are exemplary specifics of some of the pores included in the particle mask (810), the pore_001 (812) partial region in the particle mask (810) can be extracted as individual pore mask_001 (822), the pore_002 (814) partial region can be extracted as individual pore mask_002 (824), and the pore_003 (816) partial region can be extracted as individual pore mask_003 (826).

[0165] In this case, in the particle mask (810), the partial image corresponding to void_001 (812) and the image included in the individual void mask_001 (822), the partial image corresponding to void_002 (814) and the image included in the individual void mask_002 (824), and the partial image corresponding to void_003 (816) and the image included in the individual void mask_003 (826) can each be generated to include an image with the same content.

[0166] FIGS. 9 and FIGS. 10 are drawings for explaining a configuration for performing void information generation using a void information generation device according to one embodiment.

[0167] FIG. 9 is a drawing for exemplarily illustrating a configuration for generating void information according to one embodiment.

[0168] A pore information generating device according to one embodiment can generate pore information based on individual pore masks. For example, the pore information generating device can generate individual pore location information and individual pore size information for each individual pore mask, and can generate pore ratio information and pore distribution information based on a particle mask and individual pore masks.

[0169] Referring to FIG. 9, a pore information generating device according to one embodiment can generate pore distribution information regarding the distribution of pores within a particle mask (910).

[0170] For example, the pore information generating device can generate pore distribution information for the particle mask (910) in an area within the particle maximum radius (912). In this case, the number of pores can be counted according to the size of the radius, and specifically, a method of counting the number of pores within a count area (920) with a count radius (922) can be used.

[0171] For example, by increasing the count radius (922) from 0 to the particle maximum radius (912), the number of voids in each count area (920) can be counted, and void distribution information regarding the void distribution by radius size can be generated based on the number of counted voids.

[0172] However, in this case, void distribution information can be generated based on the difference in the number of voids for each region. For example, the number of voids in a region with a radius of 50 can be determined by subtracting the number of voids in a count region with a radius of 49 from the number of voids in a count region with a radius of 51.

[0173] FIG. 10 is a graph illustrating, for example, a configuration for performing void information visualization according to one embodiment.

[0174] A void information generating device according to one embodiment can generate visualization information based on void distribution information. In this case, the visualization information may include histogram information in the form of a histogram.

[0175] For example, a pore information generating device can generate histogram information using pore distribution information generated based on the number of pores counted by radius size within a particle mask.

[0176] Referring to FIG. 10, a void information generating device according to one embodiment can generate histogram information with the X-axis as the radius and the Y-axis as the number of voids.

[0177] For example, in FIG. 10, the number of voids corresponding to the first radius (1010) is shown to be around 50, and the number of voids corresponding to the second radius (1020) is shown to be over 200. In this way, the histogram information may include information showing the number of voids for each radius in the form of a histogram.

[0178] Below, a method for generating void information using a void information generating device (100) capable of performing all the aforementioned contents in the present disclosure is described. Content that overlaps with the above description may be omitted depending on the case, but all of the following methods may also be applied.

[0179] FIG. 11 is a flowchart relating to a method for generating void information according to one embodiment.

[0180] Referring to FIG. 11, a method for generating void information according to one embodiment may include an image preprocessing step (S1110), an image segmentation step (S1120), and an information generation step (S1130).

[0181] For example, a method for generating pore information may include an image preprocessing step of generating a particle mask based on an image of a cathode material cross-section including at least one particle and an image of a plurality of pores included in the particle, and performing image restoration preprocessing based on the particle mask; an image segmentation step of extracting individual pore masks for each pore based on the image of the cathode material cross-section and the particle mask; and an information generation step of generating pore information for the pores based on the individual pore masks.

[0182] The image preprocessing step (S1110) may include generating a particle mask based on an anode material cross-sectional image including an image of at least one particle and a plurality of pores included in the particle, and performing image restoration preprocessing based on the particle mask.

[0183] For example, the image preprocessing step (S1110) may include setting guide coordinates based on a cross-sectional image of the cathode material, determining particle center coordinates based on the guide coordinates using a preset SAM model, and generating a particle mask based on the particle center coordinates.

[0184] For example, the image preprocessing step (S1110) may include performing image restoration preprocessing based on a preset VSNR algorithm and energy function for the particle mask.

[0185] For example, the VSNR algorithm may include setting a particle mask as the original image and generating a reconstructed image based on an energy function and the original image. For example, the energy function may include a data term set based on the difference between the original image and the reconstructed image and a smoothing term set based on the gradient of the reconstructed image.

[0186] The image segmentation step (S1120) may include extracting individual pore masks for each pore based on the cross-sectional image of the cathode material and the particle mask.

[0187] For example, the image segmentation step (S1120) may include performing Gaussian smoothing on the particle mask. For example, the image segmentation step (S1120) may include performing threshold processing to binarize the particle mask based on a preset void threshold. For example, the image segmentation step (S1120) may include performing pixel connectivity analysis to recognize pixels connected to each other within the particle mask, and determining whether to extract each individual void mask based on the analysis results.

[0188] The information generation step (S1130) may include generating void information for the voids based on individual void masks.

[0189] For example, the information generation step (S1130) may include generating individual pore location information and individual pore size information for each individual pore mask based on individual pore masks. For example, the information generation step (S1130) may include generating pore ratio information and pore distribution information based on particle masks and individual pore masks.

[0190] FIG. 12 is a flowchart relating to an image preprocessing step according to the present disclosure.

[0191] Referring to FIG. 12, an image preprocessing step (S1110) according to one embodiment may include a guide coordinate setting step (S1210), a particle center coordinate determination step (S1220), a particle mask generation step (S1230), and an image restoration step (S1240).

[0192] The guide coordinate setting step (S1210) may include setting guide coordinates based on a cross-sectional image of the cathode material. For example, the coordinates of the center point of the cross-sectional image of the cathode material may be set as guide coordinates.

[0193] The particle center coordinate determination step (S1220) may include determining the particle center coordinates based on guide coordinates using a preset SAM model.

[0194] In the particle mask generation step (S1230), a particle mask can be generated based on the particle center coordinates. For example, a particle mask can be generated by extracting a portion corresponding to the area containing the central particle from a cross-sectional image of the cathode material.

[0195] The image restoration step (S1240) may include generating a restored image by performing image restoration preprocessing to remove curtain effects and contamination, etc., on at least one of the cathode material cross-sectional image and the particle mask.

[0196] FIG. 13 is a flowchart relating to an image segmentation step according to one embodiment.

[0197] Referring to FIG. 13, an image segmentation step (S1120) according to one embodiment may include a Gaussian smoothing step (S1310), a threshold processing step (S1320), and a pixel connectivity analysis step (S1330).

[0198] In the Gaussian smoothing step (S1310), Gaussian smoothing can be performed on the particle mask. For example, in the Gaussian smoothing step (S1310), noise remaining in the particle mask after preprocessing can be removed using Gaussian smoothing.

[0199] The threshold processing step (S1320) may include performing threshold processing to binarize the particle mask based on a preset void threshold. In some cases, the threshold processing step (S1320) may perform local threshold processing by setting different void thresholds for each region of the particle mask.

[0200] The pixel connectivity analysis step (S1330) may include performing pixel connectivity analysis to recognize pixels connected to each other within the particle mask and group them by object, and determining whether to extract each individual void mask based on the analysis results.

[0201] FIG. 14 is a flowchart relating to an information generation step according to one embodiment.

[0202] Referring to FIG. 14, an information generation step (S1130) according to one embodiment may include an individual pore information generation step (S1410), a total pore information generation step (S1420), and a pore information visualization step (S1430).

[0203] The individual void information generation step (S1410) may include generating individual void information based on individual void masks. In this case, the individual void information may include individual void location information and individual void size information.

[0204] For example, individual pore location information may include information regarding the location of each pore existing within the particle mask. For example, individual pore size information may include information regarding the size of each pore existing within the particle mask.

[0205] The step of generating total void information (S1420) may include generating total void information based on individual void information. In this case, the total void information may include void ratio information and void distribution information.

[0206] For example, the pore ratio information may include information regarding the ratio of each pore classified according to preset criteria. For example, the pore count distribution information may include information regarding the number of individual pores distributed according to the distance to each pore relative to the central particle of the particle mask.

[0207] In the pore information visualization step (S1430), the visualization information may include information that visualizes the distribution of pores included in the particle mask. For example, it may include graph information regarding the pore distribution, scatter plot information showing the pore distribution scattered in the form of points, histogram information regarding the pore distribution, etc.

[0208] For example, in the void information visualization step (S1430), histogram information regarding a histogram can be generated with the distance to each void relative to the central particle of the particle mask as the X-axis and the number of individual voids for each distance as the Y-axis.

[0209] For example, in the void information visualization step (S1430), histogram information can be generated based on void size distribution information. For example, after classifying void sets by the size of each void, histogram information can be generated for a histogram in which each size set is on the X-axis and the number of individual voids for each size set is on the Y-axis.

[0210] As described above, the present disclosure can provide a void information generating apparatus and method capable of generating information regarding voids based on a given image.

[0211] The devices, methods, configurations, glyphs, and operations described herein may be implemented in digital electronic circuits, or computer software, firmware, or hardware comprising structures disclosed herein and structural equivalents, or combinations of one or more of these. The glyphs described herein may be implemented as one or more computer programs, for example, as one or more modules of computer program instructions encoded on a computer storage medium to control execution by a data processing device or operation by a data processing device. Program instructions may be encoded in artificially generated propagated signals, for example, mechanically generated electrical, optical, or electromagnetic signals generated to encode information for transmission to a suitable receiver device for execution by a data processing device. The computer storage medium may be or may include a computer-readable storage device, a computer-readable storage substrate, a random or serial access memory array or device, or a combination of one or more of these. Although the computer storage medium is not a propagated signal, the computer storage medium may be a source or destination of computer program instructions encoded in an artificially generated propagated signal. Additionally, a computer storage medium may be one or more individual physical components or media (e.g., multiple CDs, disks, or other storage devices) or may include. The operations described herein may be implemented as operations performed by a data processing device on data stored in one or more computer-readable storage devices or data received from other sources.

[0212] The foregoing description is merely an illustrative explanation of the technical concept of the present disclosure, and those skilled in the art to which the present disclosure pertains may make various modifications and variations within the scope of the essential characteristics of the technical concept. Furthermore, since these embodiments are intended to explain rather than limit the technical concept of the present disclosure, the scope of the technical concept is not limited by these embodiments.

[0213]

[0214] CROSS-REFERENCE TO RELATED APPLICATION

[0215] This patent application claims priority pursuant to Section 119(a) of the U.S. Patent Act (35 USC § 119(a)) to Korean Patent Application No. 10-2024-0161566 filed on November 14, 2024, all of which are incorporated by reference into this patent application. Furthermore, this patent application claims priority in countries other than the United States for the same reasons as above, all of which are incorporated by reference into this patent application.

Claims

1. A preprocessing unit that generates a particle mask based on a cross-sectional image of an anode material including an image of at least one particle and a plurality of pores included in said particle, and performs image restoration preprocessing based on said particle mask; An image segmentation unit that extracts individual pore masks for each pore based on the cross-sectional image of the anode material and the particle mask; A void information generating device comprising an information generating unit that generates void information for the voids based on the individual void masks.

2. In Paragraph 1, The above preprocessing unit is, A void information generating device that sets guide coordinates based on the cross-sectional image of the anode material, determines particle center coordinates based on the guide coordinates using a preset SAM model, and generates the particle mask based on the particle center coordinates.

3. In Paragraph 1, The above preprocessing unit is, A void information generating device that performs image restoration preprocessing based on a preset VSNR algorithm and energy function for the particle mask.

4. In Paragraph 3, The above VSNR algorithm is, The method includes setting the particle mask as the original image and generating a reconstructed image based on the energy function and the original image. The above energy function is, A void information generating device comprising a data term set based on the difference between the original image and the restored image, and a smoothing term set based on the gradient of the restored image.

5. In Paragraph 3, The above preprocessing unit is, A void information generation device that performs image restoration preprocessing in a direction in which the applied value of the energy function decreases using a preset Euler-Lagrange equation and stochastic gradient descent (SGD).

6. In Paragraph 1, The above image segmentation unit is, A void information generating device that performs Gaussian smoothing on the above particle mask.

7. In Paragraph 1, The above image segmentation unit is, A void information generating device that performs threshold processing to binarize the particle mask based on a preset void threshold.

8. In Paragraph 1, The above image segmentation unit is, A void information generating device that performs pixel connectivity analysis to recognize interconnected pixels within the particle mask and determines whether to extract each individual void mask based on the analysis results.

9. In Paragraph 1, The above information generation unit is, A void information generating device that generates individual void location information and individual void size information for each individual void mask based on the individual void mask.

10. In Paragraph 1, The above information generation unit is, A pore information generating device that generates pore ratio information and pore distribution information based on the particle mask and the individual pore mask.

11. In Paragraph 10, The above void distribution information is, Includes pore number distribution information and pore size distribution information, The above information generation unit is, A void information generating device that generates histogram information based on at least one of the above void number distribution information and the above void size distribution information.

12. An image preprocessing step of generating a particle mask based on a cross-sectional image of an anode material comprising an image of at least one particle and a plurality of pores included in said particle, and performing image restoration preprocessing based on said particle mask; An image segmentation step for extracting individual pore masks for each pore based on the cross-sectional image of the anode material and the particle mask; A method for generating void information comprising an information generation step of generating void information for the voids based on the individual void masks above.

13. In Paragraph 12, The above image preprocessing step is, A method for generating void information comprising setting guide coordinates based on the cross-sectional image of the anode material, determining particle center coordinates based on the guide coordinates using a preset SAM model, and generating the particle mask based on the particle center coordinates.

14. In Paragraph 12, The above image preprocessing step is, A method for generating pore information comprising performing image restoration preprocessing based on a preset VSNR algorithm and energy function for the particle mask.

15. In Paragraph 14, The above VSNR algorithm is, The method includes setting the particle mask as the original image and generating a reconstructed image based on the energy function and the original image. The above energy function is, A method for generating void information comprising a data term set based on the difference between the original image and the restored image, and a smoothing term set based on the gradient of the restored image.

16. In Paragraph 14, The above image preprocessing step is, A method for generating void information comprising performing image restoration preprocessing in a direction in which the applied value of the energy function decreases using a preset Euler-Lagrange equation and stochastic gradient descent (SGD).

17. In Paragraph 12, The above image segmentation step is, A method for generating void information including performing Gaussian smoothing on the above particle mask.

18. In Paragraph 12, The above image segmentation step is, A method for generating void information comprising performing threshold processing to binarize the particle mask based on a preset void threshold.

19. In Paragraph 12, The above image segmentation step is, A method for generating void information comprising performing a pixel connectivity analysis that recognizes pixels connected to each other within the particle mask and determining whether to extract each individual void mask based on the analysis results.

20. In Paragraph 12, The above information generation step is, A method for generating void information, comprising generating individual void location information and individual void size information for each individual void mask based on the individual void mask.

21. In Paragraph 12, The above information generation step is, A method for generating pore information, comprising generating pore ratio information and pore distribution information based on the particle mask and the individual pore mask.

Images