Systems, devices and methods for three-dimensional analysis of carbon black
A 3D reconstruction method for carbon black aggregates using TEM images and sphere-based optimization addresses the challenges of polydispersity and computation time, providing detailed morphological insights for improved analysis.
Patent Information
- Application Number
- JP2024518648
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2021-09-23
- Filing Date
- 2022-09-23
- Publication Date
- 2024-11-01
AI Technical Summary
Existing techniques for three-dimensional (3D) characterization of carbon black morphology are hindered by factors such as polydispersity, long computation times, and ambiguous aggregate surfaces, limiting statistically significant analysis.
A computer-implemented method using two-dimensional (2D) transmission electron microscopy (TEM) images to identify carbon black aggregates, and a 3D agglomerate model is developed by solving an optimization problem for the size and position of spheres within a defined volume, enabling efficient 3D reconstruction.
The method allows for detailed, statistically relevant 3D reconstruction of carbon black aggregates, reducing computation time and providing accurate morphological properties, such as specific surface area and void volume, enhancing understanding of carbon black interactions with rubber.
Smart Images

Figure 2024540704000001_ABST
Abstract
Description
[Technical field]
[0001] CROSS-REFERENCE TO RELATED APPLICATIONS This application claims the benefit of and priority to U.S. Provisional Patent Application No. 63 / 247,327, filed September 23, 2021, the entire contents of which are incorporated herein by reference in their entirety. [Background technology]
[0002] Two-dimensional (2D) transmission electron microscopy (TEM) has enabled the linking of carbon black morphology with application properties. Similarly, electron tomography-based three-dimensional (3D) aggregate reconstructions have revealed aggregate mass distributions. However, 3D characterization of carbon black has been hindered by a variety of factors, some of which relate to the characteristics of the carbon black and others to the computational modeling involved in existing attempts to characterize carbon black morphology. For example, the polydispersity of carbon black, the very long elapsed time typically required for computational modeling of carbon black, and the fuzzy aggregate surfaces generated in 3D aggregate reconstructions have hindered accurate and statistically significant 3D characterization of carbon black morphology. Thus, much remains to be improved in techniques for computational 3D analysis of carbon black morphology using electron microscopy images of carbon black aggregates. Summary of the Invention
[0003] It is to be understood that both the following general description and the following detailed description are exemplary and explanatory only and are not restrictive.
[0004] An embodiment of the present disclosure includes a computer-implemented method that includes identifying carbon black agglomerates using a two-dimensional (2D) transmission electron microscope (TEM) image and determining a number of spheres based at least on an approximation to a volume of the carbon black agglomerates. The computer-implemented method also includes generating a three-dimensional (3D) agglomerate model of the carbon black agglomerates by determining a solution to an optimization problem for an objective function based on a size of the spheres and a position of the spheres within the defined volume, the solution representing the 3D agglomerate model.
[0005] Additional elements and advantages of the present disclosure will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the present disclosure. The advantages of the present disclosure may be attained by means of the elements and combinations particularly pointed out in the appended claims.
[0006] This summary is not intended to identify key or essential features of the disclosure, but merely to summarize certain features and variations thereof. Additional details and features are described in the following sections. Moreover, both the foregoing summary and the following detailed description are exemplary and explanatory only and are not intended to limit the embodiments of the present disclosure.
[0007] The accompanying drawings are an integral part of the present disclosure and are incorporated herein. The drawings illustrate exemplary embodiments of the present disclosure and, together with the description and claims, serve to at least partially explain various principles, elements, or aspects of the present disclosure. The embodiments of the present disclosure are described more fully below with reference to the accompanying drawings. However, various elements of the present disclosure may be implemented in many different forms and should not be construed as being limited to the implementations set forth herein. Like numbers refer to like elements throughout. [Brief description of the drawings]
[0008] [Figure 1]1 illustrates an example process flow for 3D analysis of carbon black morphology in accordance with one or more embodiments of the present disclosure. [Diagram 2] FIG. 1 illustrates an example of an operating environment (including a computing system) for 3D analysis of carbon black morphology in accordance with one or more embodiments of the present disclosure. The operating environment includes an electron microscope system and a computing system. [Diagram 3] Tilted TEM images of multiple carbon black aggregates, with 63 carbon black aggregates labeled with their respective numbers. Thirty selected aggregates are marked with their respective closed curves to indicate that they belong to the reconstruction and analysis group. [Figure 4] 13 illustrates an example of determining a goodness-of-fit metric for a cost function at a projection angle of 0° for a reconstructed aggregate, in accordance with one or more embodiments of the present disclosure. [Diagram 5] 1 is a photographic projection of a 3D printed assembly of 62 spheres mounted on a rotating shaft according to one or more embodiments of the present disclosure. The 3D printed assembly represents a model of carbon black aggregates at a macroscopic scale. [Figure 6] FIG. 6 illustrates a 3D reconstruction of the 3D printed assembly shown in FIG. 5 in accordance with one or more embodiments of the present disclosure. [Figure 7A-C] [Figure 7A] Non-tilted (0°) TEM micrograph of a carbon black aggregate. [Figure 7B] Non-tilted tomogram of the carbon black aggregate shown in Figure 7A as determined using conventional TEM tomography reconstruction. [Figure 7C] Non-tilted sphere-based 3D reconstruction of a carbon black aggregate according to an embodiment of the present disclosure. [Figure 8] 13 illustrates the dependence of the average reconstruction objective function on the tilt angle interval and tilt angle range for the reconstruction of 30 N330 carbon black aggregates, in accordance with one or more embodiments of the present disclosure. [Figure 9]Shown are zero-tilt (0°) TEM images and 0° and 90° 3D aggregate model images of seven of the 30 selected carbon black aggregates shown in Figure 1. Images are not to scale. [Figure 10] 13 shows the distribution of converged average objective function values for 265 3D reconstructions for each N330 carbon black aggregate, in accordance with one or more embodiments of the present disclosure. [Figure 11] 1 shows the distribution of aggregate reconstitution times for 256 carbon black aggregates in accordance with one or more embodiments of the present disclosure. [Figure 12] 3 shows the distribution of 3D specific surface areas (3DSSA) calculated from 256 3D reconstructions for each N330 carbon black aggregate, according to one or more embodiments of the present disclosure. The subset of 256 3D reconstructions corresponds to the reconstructions of the selected 30 N330 carbon black aggregates marked in FIG. [Figure 13] 3 shows the inequality metric (referred to as "relative Z") as a function of relative 3D void volume (3DVV) for reconstructions of 30 N330 carbon black aggregates, in accordance with one or more embodiments of the present disclosure. Each reconstruction corresponds to a respective one of the 30 N330 carbon black aggregates marked in FIG. [Figure 14] 3 shows 3D aggregate volume (from voxel count) versus 2D aggregate volume (from (8 / 3)A2 / P) for reconstructions of 30 N330 carbon black aggregates, according to one or more embodiments of the present disclosure, each corresponding to a respective one of the 30 N330 carbon black aggregates marked in FIG. [Figure 15] 3 shows the distribution of specific surface areas of 30 N330 carbon black aggregate reconstructions, each corresponding to a respective one of the 30 N330 carbon black aggregates marked in FIG. [Figure 16]3 shows the relationship between 3DVV and 2DV' / V for the reconstruction of 30 N330 carbon black aggregates, in accordance with one or more embodiments of the present disclosure, each reconstruction corresponding to a respective one of the 30 N330 carbon black aggregates marked in FIG. [Figure 17] 1 illustrates an example process flow for 3D analysis and identification of carbon black, in accordance with one or more embodiments of the present disclosure. [Figure 18] 1 illustrates an example of a system for generating a predictive model configured to identify a type of carbon black aggregate, according to one or more embodiments described herein. [Figure 19] 1 illustrates an example operating environment (including a computing system) for designating carbon black aggregates as being a particular carbon black, according to one or more embodiments described herein. [Figure 20] 1 illustrates an example of a method according to one or more embodiments of the present disclosure. [Figure 21] 1 illustrates another example of a method according to one or more embodiments of the present disclosure. [Figure 22] 1 illustrates another example of a method according to one or more embodiments of the present disclosure. [Figure 23] 13 illustrates yet another example of a method according to one or more embodiments of the present disclosure. [Figure 24] 1 illustrates another example of a method according to one or more embodiments of the present disclosure. [Diagram 25] 1 illustrates an example of a computing system for implementing 3D reconstruction and analysis of carbon black aggregates in accordance with one or more embodiments of the present disclosure. [Figure 26] 1 illustrates another example of a computing system for implementing 3D analysis of carbon black aggregates in accordance with one or more embodiments of the present disclosure. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0009] The present disclosure recognizes and addresses, among other technical challenges, the problem of three-dimensional reconstruction and analysis of carbon black morphology using electron microscopy images of carbon black. The embodiments of the present disclosure, individually or in combination, advance the TEM-based elucidation of carbon black aggregate morphology by enabling direct modeling of the mass distribution of carbon black aggregates in a statistically relevant manner and at time scales suitable for widespread laboratory applications. More specifically, the embodiments of the present disclosure provide computing systems, computing devices, computer program products, and computer-implemented methods for sphere-based tomographic reconstruction and analysis of carbon black aggregates. More specifically, the embodiments of the present disclosure provide a 3D reconstruction approach that relies on solving an optimization problem for an objective function based on the size and position of multiple spheres within a defined volume. Such an approach results in a 3D aggregate model that can be compactly defined in terms of a 4-tuple for each sphere in the model, the tuple including a 3D position vector and a parameter that defines the size of the sphere. The embodiments of the present disclosure, individually or in combination, can be applied to binarized TEM images of carbon black (CB) aggregates, resulting in a complete description of the spatial distribution of the aggregates' mass. Although embodiments of the present disclosure are illustrated with respect to N330 CB aggregates, the disclosure is not limited in that respect. Indeed, the principles and practical applications of the present disclosure relating to 3D reconstruction and analysis of carbon black morphology can be directed to other types of carbon black and their aggregates, such as N110 carbon black, N220 carbon black, N330 carbon black, N550 carbon black, N660 carbon black, N772 carbon black, N880 carbon black, and N990 carbon black.
[0010] The embodiments of the present disclosure result in a 3D aggregate model that allows for the determination of values of multiple morphological properties of carbon black aggregates. Not only do the morphological property values match existing 2D analyses of carbon black, but the availability of the location and size of the carbonaceous particles that form the carbon black aggregates allows for a more detailed analysis than conventional tomographic reconstructions.
[0011] The disclosed embodiments are illustrated with a fixed number of spheres per carbon black aggregate and a 3D reconstruction based on binary TEM images. However, the disclosed 3D sphere-based approach to carbon black aggregate reconstruction is not limited in that respect. Indeed, in some embodiments, the 3D reconstruction of the carbon black aggregate can be based on the number of spheres treated as an optimization parameter, and thus dynamically changes during the 3D reconstruction. Additionally, or in other embodiments, rather than using binary TEM images and corresponding modeled binary projections, the disclosed 3D sphere-based approach can be based on grayscale TEM images and corresponding simulated grayscale projections. Such modeled grayscale projections can be generated based on appropriate estimators of electron transmission intensity as a function of mass and / or gain / offset adjustments. In some cases, the estimators can be based on a parameterized decaying exponential function, and the parameters defining the decay function can be used as optimization parameters in the 3D sphere-based reconstruction.
[0012] Furthermore, although the embodiments of the present disclosure are illustrated with reference to spheres and carbon black, the principles and practical applications of the present disclosure are not limited to any one of them. Indeed, the mechanisms underlying the imaging and approach to 3D reconstruction described in this disclosure can also be applied to other structures of carbon, organic elements, or inorganic elements, using geometric objects of non-spherical geometry in modeling those other structures in 3D. The approach to imaging and 3D reconstruction can ultimately be applied to materials of any composition, where the material has one or more structures that can be modeled by a group of geometric objects. As an example, a curved tubular structure such as a single-walled carbon nanotube (CNT) or multi-walled CNT can be modeled as an assembly of right circular cylinders, each of which is described by seven parameters, namely, three parameters indicating a first position vector r corresponding to a first circular base of the right circular cylinder, three other parameters indicating the position r' of a second circular base of the right circular cylinder, and a single parameter indicating the size of the first and second circular bases. In some cases, such curved tubular structures can be modeled as an assembly of truncated cones, each of which is described by eight parameters: three parameters indicating a first position vector r corresponding to a first circular base of the truncated cone, three other parameters indicating the position r' of a second circular base of the right circular cylinder, a parameter indicating the size of the first circular base, and a parameter indicating the size of the second circular base. The approach to 3D reconstruction according to the present disclosure is not limited to such non-spherical geometric objects, and objects having other geometric shapes are also contemplated.
[0013] Referring to the drawings, Figure 1 illustrates an example of a process flow 100 for 3D analysis of carbon black morphology in accordance with one or more embodiments of the present disclosure. Process flow 100 includes multiple stages, represented by respective blocks in Figure 1, which may be implemented individually or in combination with one another. Process flow 100 may be implemented, at least in part, in an operating environment 200 illustrated in Figure 2.
[0014] TEM-based tomography involves radiation penetrating the mass at multiple tilt angles around an axis of rotation contained in the cross-sectional plane of the mass, generating projection images containing information on that cross-sectional plane. Computed tomography then involves processing these projection images to create a 3D model of the mass under investigation. The resulting data is referred to as a tomogram, and the 3D model can be defined as a density function of 3D vector positions (ρ(r)=ρ(x,y,z)). Thus, the process flow 100 includes an image acquisition stage 110 that can access imaging data that defines a TEM image of the carbon black aggregate. These TEM images include a series of TEM 2D images of the carbon black aggregate, each TEM 2D image generated at a respective tilt angle. Each TEM 2D image in the series may be referred to as a TEM tilt image.
[0015] As shown in FIG. 2, a series of TEM tilt images of the carbon black aggregates can be generated by an imaging system 205. The imaging system 205 includes an electron microscope instrument 210 that can generate an electrical signal representative of the distribution of the carbon agglomerates dispersed on the carbon film. The carbon agglomerates can be ultrasonically dispersed in a solvent to form a plurality of carbon black aggregates that can then be deposited on the carbon film (e.g., a glow discharge activated carbon film). To that end, the electron microscope instrument 210 includes a source device that serves as a source of electrons that can be injected into a collimator device. The collimator device can form a beam of electrons having approximately the same kinetic energy. Such a beam may be referred to as a monochromatic electron beam. The collimator device can be elongated along the incidence direction (indicated by z in FIG. 2). From the output of the collimator device, the monochromatic electron beam can enter a focusing device in the electron microscope instrument 210. The focusing device can direct the monochromatic electron beam to a sample holder device that is also part of the electron microscope instrument 210.
[0016] The sample holder arrangement may include a TEM grid with a carbon film attached thereto. The sample holder arrangement may be embodied with a goniometric stage (single or multi-axis). The sample containing the carbon film may thus be rotated over a defined tilt angle range [-Θ,Θ] with defined successive tilt angle intervals Δθ. As a result of such rotation, the sample is sequentially positioned at a series of tilt angles over the defined tilt angle range. At each tilt angle, the sample may be subjected to a monochromatic electron beam. In some cases, Θ=70° and Δθ=2°, resulting in a series of 71 tilt angles. In some cases, Θ=70° and Δθ=10°, resulting in a series of 15 tilt angles. The control unit 215 may rotate the goniometric stage over a defined tilt angle range [-Θ,Θ] with defined successive tilt angle intervals Δθ. At each tilt angle, the control unit 215 may also apply a monochromatic electron beam to the sample. To cause such rotation and exposure to the monochromatic electron beam, the control unit 215 may execute a control sequence having processor-executable instructions that, in response to being executed, cause the goniometric stage to rotate and the monochromatic electron beam to impinge on the sample. The control sequence may have further processor-executable instructions that, in response to being executed, cause a focusing device to focus (or refocus) the electron beam before collecting each image.
[0017] The electron microscope instrument 210 may also include a sensor device capable of detecting the scattered electrons that have passed through the TEM grid. In response to such detection, the sensor device may generate an electrical signal representative of the arrangement of the carbon agglomerates dispersed within the carbon film. The electrical signals may be generated sequentially according to a series of tilt angles within a defined tilt angle range [-Θ,Θ], i.e., an electrical signal is generated for each tilt angle in the series. As a result, a batch of electrical signals may be generated for each tilt angle in the series of tilt angles.
[0018] Batches of electrical signals can be individually passed from the electron microscope instrument 210 to the imaging subsystem 220 via a bus architecture. The imaging subsystem 220 can generate imaging data by processing batches of electrical signals generated at tilt angles over a range of tilt angles, for which the imaging data defines a TEM tilt image of the carbon black aggregates dispersed within the carbon film. By processing batches of electrical signals generated over a range of tilt angles, the imaging subsystem 220 can generate imaging data defining a series of TEM 2D images of the carbon black aggregates generated at each tilt angle. The generation of such imaging data can be automated by automating the rotation of the goniometric stage and subsequent exposure to the monochromatic electron beam via the control unit 215. The automated generation of imaging data is sometimes referred to as automated image acquisition.
[0019] The imaging subsystem 220 may then persist the imaging data in one or more memory devices 230 (referred to as data storage 230). The imaging data may be persisted in one or more files in a file system configured in the data storage 230. Each of the one or more files may be formatted according to an appropriate format for image storage.
[0020] As an example, automated image acquisition for a series of tilt angles spanning the range [-70°, 70°] at Δθ=10° intervals can take place in a fraction of an hour (e.g., about 20 to about 40 minutes). The time period elapsed in such automated image acquisition can include short time intervals elapsed for each image and longer time intervals interspersed between the short intervals. The longer time intervals correspond to respective set-up periods prior to acquiring imaging data at a particular tilt angle. In some cases, each of the short intervals can span about 30 seconds and each of the longer time intervals can span several minutes. Thus, automated image acquisition can be substantially faster than existing techniques for 2D analysis of carbon black. Thus, embodiments of the present disclosure can efficiently acquire and store many Δθ=10° TEM tilt image series. Acquisition and storage of multiple TEM tilt image series can yield a wealth of unaltered, reusable imaging data that can be easily applied to any current or future reconstruction approach.
[0021] For illustrative purposes only, a N330 carbon black sample can be ultrasonically dispersed in chloroform and deposited on a glow discharge activated carbon film grid. A series of TEM tilt images of the N330 carbon black aggregates can be generated using the electron microscope instrument 210. To that end, the goniometric stage of the electron microscope instrument 210 can rotate the sample over a range of [-70°, 70°] at defined successive tilt angle intervals Δθ (e.g., 2° or 10°). Furthermore, at each tilt angle, the sample can be subjected to a monochromatic electron beam of 120 kV. Furthermore, as described above, the electrical signals generated by the sensor device of the electron microscope instrument 210 can be used to generate imaging data. The sensor device can include a charge-coupled device (CCD) having a 2672×4008 pixel array. The imaging resolution can be configured to be about 20 pixels per average particle diameter, which corresponds to about 1.67 nm per pixel for N330 carbon black. Its imaging resolution is approximately twice the 10 pixels per average particle diameter typically used for 2D measurements of carbon black. A representative aggregate field containing 50-100 aggregates can be selected for tomographic imaging and reconstruction according to the present disclosure. Of the 63 identified reconstructable aggregates in the initially acquired series, 30 were selected for initial reconstruction and method evaluation. As an example, FIG. 3 is a TEM tilt image corresponding to a tilt angle of 0° (referred to as zero tilt) in the series of TEM tilt images. The selected 30 carbon black aggregates used to illustrate various reconstruction and analysis aspects of the present disclosure are individually marked in FIG. 3 with either a circle or an ellipse. The selected 30 carbon black aggregates are a subset of the multiple reconstructable aggregates present in the series of TEM tilt images. Here, reconstructable aggregates refer to carbon black aggregates that are visible in all images of the series of TEM tilt images. The selected carbon aggregates may be reconfigurable aggregates that satisfy a non-occlusion criterion indicating that the reconfigurable aggregate does not occlude or is not occluded by another reconfigurable aggregate in the field of view at any tilt angle through the defined tilt angle range [-Θ,Θ].
[0022] The generation of the TEM 2D images is part of the image acquisition stage 110. Another part of the image acquisition stage 110 includes the transfer of imaging data from the imaging system 205 and can be implemented by the modeling system 245. More specifically, a computing device 250 that is part of the modeling system 245 can access the series of TEM 2D images 234 of the carbon black aggregates generated at each tilt angle. To that end, the computing device 250 can include an acquisition module 252 that can access imaging data defining the series of TEM 2D images from the data storage 230. Accessing the imaging data can include programmatically downloading the series of TEM 2D images from the data storage 230. For example, the acquisition module 252 can download the TEM 2D images via an application programming interface (API). By executing one or more function calls of the API, the acquisition module 252 can receive image files 234 corresponding to the series of TEM 2D images and can retain the files 234 in one or more non-volatile memory devices 278 (referred to as memory 278). As another example, the ingestion module 252 can download the TEM 2D by executing a script to copy the image file 234 from a file system in the data storage 230 to the memory 278. The memory 278 can be integrated into the computing device 250. The computing device 250 can include computing resources (not shown) comprising, for example, a central processing unit (CPU), a graphics processing unit (GPU), a tensor processing unit (TPU), memory, disk space, input bandwidth and / or output bandwidth, interface(s) (such as an I / O interface or API, or both), controller device(s), a power source, combinations of the foregoing, and / or similar resources.
[0023] Each TEM tilt image of the carbon black aggregate depicts a plurality of carbon black aggregates. As shown in FIG. 3, the depicted carbon black aggregates can be well separated and easily distinguished from one another. The process flow 100 can include an aggregate image processing stage 120 that enables the TEM tilt image to be segmented into a plurality of TEM tilt images depicting respective carbon black aggregates. Thus, for a series of TEM tilt images, the aggregate image processing stage 120 can enable identification of individual carbon black aggregates within the series of TEM tilt images and generation of a plurality of other series of TEM tilt images, each of which corresponds to one respective carbon black aggregate. In other words, an implementation of the image processing stage 120 may result in a plurality of series of TEM tilt images for the Q carbon black aggregates, where a first series of TEM tilt images of the plurality of series of TEM tilt images corresponds to a first carbon black of the Q carbon black aggregates, a second series of TEM tilt images of the plurality of series of TEM tilt images corresponds to a second carbon black of the Q carbon black aggregates, continuing through a j series of TEM tilt images of the plurality of series of TEM tilt images for a j series of carbon black aggregates of the Q carbon black aggregates, and further continuing through a Q series of TEM tilt images of the plurality of series of TEM tilt images for a Q series of carbon black aggregates of the Q carbon black aggregates. By way of example only, for carbon black aggregate No. 21 of FIG. 3, an implementation of the aggregate image processing stage 120 may provide a series of TEM tilt images for that particular carbon black aggregate. Continuing further with the example, for carbon black aggregate No. 30 of FIG. 3, an implementation of the aggregate image processing stage 120 may provide a series of TEM tilt images for that particular carbon black aggregate.
[0024] The modeling system 245 may also implement the aggregate image processing stage 120 via a computing device 250 (FIG. 2). The computing device 250 may include a segmentation module 256 that may identify a particular carbon aggregate in the TEM tilt image and then extract a section of the TEM tilt image depicting the particular carbon aggregate and, optionally, a portion of the image background. The portion of the image background corresponds to an area of the TEM tilt image surrounding the carbon black aggregate and lacking carbon mass. The segmentation module 256 may identify the particular carbon black aggregate by applying one or more machine vision techniques to determine pixels indicative of a perimeter of the particular carbon black aggregate. Extracting the section of the TEM tilt image may include selecting imaging data that defines values of pixels in and including the perimeter. The imaging data may optionally define values of pixels indicative of a surrounding area of the carbon black aggregate. The machine vision technique(s) may include template matching techniques for object localization as described herein.
[0025] The segmentation module 256 can identify the particular carbon black aggregate across the series of TEM tilt images corresponding to the respective tilt angles. To that end, in some cases, the segmentation module 256 can apply a template matching technique to the series of TEM tilt images. Applying the template matching technique can include generating a template (also referred to as a feature) for the particular carbon black aggregate in a first TEM tilt image of the series of TEM tilt images. In one example, the first TEM tilt image can be a series of no-tilt TEM images such as FIG. 3. Here, for purposes of illustration, the template (or feature) of the carbon black aggregate refers to a data structure that identifies a set of pixels in a binary TEM tilt image of the carbon black aggregate, where the pixels in the set are adjacent to each other. Applying the template matching technique also includes comparing the template for the particular carbon black aggregate to pixels within a defined search area in a next TEM tilt image of the series of TEM tilt images. Continuing with this example, the next TEM tilt image can correspond to a tilt angle Δθ.
[0026] Furthermore, since the particular carbon black aggregate is the aggregate being searched for across the TEM tilt images, the defined search area may be a rectangular area having a base (or first side) that includes the location of the particular carbon black aggregate along the tilt axis in the field of view (e.g., the x-axis in FIG. 3 ) and that spans a limited portion of the entire field of view along the tilt axis. That location of the particular carbon black aggregate along the tilt axis may be the location of the particular carbon black aggregate in the first TEM tilt image. The rectangular area may span the entire length of the field of view along an axis perpendicular to the rotation axis. In FIG. 3 , the rotation axis is the x-axis and the other axis is the y-axis. By limiting the defined search area in such a manner, searching and identifying the particular carbon black aggregate in the next TEM tilt image may be computationally efficient. Comparing the pixels in the search area to the template may include comparing pixels in a binarized version of the next TEM tilt image to the template. The segmentation module 256 may identify the particular carbon black aggregate in such binarized image in response to determining that a threshold number of pixels corresponds (or matches) to the template. Other matching criteria other than a threshold number of corresponding pixels may also be contemplated. After identifying a particular carbon black aggregate in the binarized version of the next TEM tilt image, the segmentation module 256 can generate an individualized TEM tilt image for the carbon black aggregate at the tilt angle of the next TEM tilt image by cropping a portion of the next TEM tilt image, which portion depicts the carbon black aggregate.
[0027] Further, the segmentation module 256 can continue to search for the particular carbon black aggregate in subsequent TEM tilt images in the series of TEM tilt images. Continuing with the above example, the subsequent TEM tilt images can correspond to a tilt angle of 2Δθ. Because the position of the particular carbon black aggregate along the tilt axis (x-axis in FIG. 3 ) can remain unchanged upon rotation about that axis, the segmentation module 256 can limit the search to the same rectangular region used in the previous search. Rather than utilizing the template defined in the first TEM tilt image in searching for the carbon black aggregate in the subsequent TEM tilt images, the segmentation module 256 can determine the next template for the particular carbon black aggregate based on the immediately preceding TEM tilt image in which the particular carbon black aggregate was identified. That is, rather than applying a common template across the series of TEM tilt images, the segmentation module 256 can update the template for each adjacent pair of TEM tilt images. In other words, for a tilt TEM image at a defined tilt angle, the segmentation module 256 applies template matching using the template τ, and for a next tilt TEM image at a successive defined tilt angle, the segmentation module 256 updates from template τ to template τ', and then uses the updated template τ' in the next tilt TEM image. After identifying a particular carbon black aggregate in the binarized version of the subsequent tilt-next TEM image, the segmentation module 256 can generate an individualized TEM tilt image of the carbon black aggregate at a tilt angle (e.g., 2Δθ) of the subsequent tilt-next TEM image by cropping a portion of the subsequent tilt-next TEM image, which portion depicts the carbon black aggregate.
[0028] In some cases, instead of utilizing a rectangular region as a search region across the TEM tilt images in the series of TEM tilt images, further computational efficiency can be obtained by searching within another more limited rectangular region. The other limited rectangular region also includes the location of the particular carbon black aggregate along the tilt axis in the field of view (e.g., the x-axis in FIG. 3 ) and has a base (or first side) that spans a limited portion of the entire field of view along the tilt axis. Additionally, the other rectangular region can have a height (or second side) that includes the expected location of the particular carbon black aggregate on an axis perpendicular to the rotation axis for a defined tilt angle that corresponds to the TEM tilt image in which the particular carbon black aggregate is being searched. The segmentation module 256 can determine the expected location by multiplying the location of the particular carbon black aggregate in the zero-tilt TEM tilt image by a coefficient equal to the cosine of the defined tilt angle. Such a multiplication can account for the change in location of the particular carbon black aggregate in the field of view upon rotation about the tilt axis.
[0029] Other defined search areas may be utilized to search for specific carbon black aggregates. In some cases, a respective defined search area may be defined for each TEM tilt image in the series of TEM tilt images. Each one of the respective defined search areas may be defined relative to a reference carbon black aggregate that is common across the series of TEM tilt images. More specifically, the reference carbon black aggregate may be pre-identified in each of the series of TEM tilt images, for example also using template matching. Thus, for each TEM tilt image in the series of TEM tilt images, the segmentation module 256 may determine a location of the reference carbon black aggregate. The location of the reference carbon black aggregate in one TEM tilt image of the series of TEM tilt images may serve as a reference location of the defined search area in that TEM tilt image. In particular, the segmentation module 256 may configure the reference location as a geometric center of the defined search area for that TEM tilt image. Furthermore, the defined search area may be a rectangular area centered on the configured geometric center and may have a defined first side (or base) and a defined second side (or height). Thus, the location of each defined search area may vary across the series of TEM tilt images. Additionally, for each TEM tilt image in the series of TEM tilt images, the segmentation module 256 can adjust the size of the defined second side of the current image by multiplying the size of the second side in the previous image by a factor equal to the cosine of the defined tilt angle corresponding to the current image. After the segmentation module 256 defines a search area for a reference carbon black agglomerate in a TEM tilt image of the series of TEM tilt images, the segmentation module 256 can search for and identify the specific carbon black agglomerate in that TEM tilt image using template matching according to aspects described herein.
[0030] By using a search region defined relative to a reference carbon black aggregate, various computational efficiencies can be achieved. For example, defining a search region relative to a carbon black aggregate can easily account for possible variations in the position of a particular carbon black aggregate along the tilt axis (x-axis in FIG. 3) across two or more images in a series of tilt TEM images.
[0031] As a result of applying the template matching technique or other object location technique to each of the TEM tilt images in the series of TEM tilt images, the segmentation module 256 can generate image data defining an individualized TEM tilt image corresponding to a section containing a particular carbon black aggregate. As a result, the segmentation module 256 can generate a series of individualized TEM tilt images corresponding to respective tilt angles and depicting the particular carbon black aggregate separately from other carbon black aggregates in the TEM tilt images. The imaging data defining the series of individualized TEM tilt images can be retained in memory 278. The segmentation module 256 can repeat the application of the template matching technique described herein to identify other carbon black aggregates across the series of TEM tilt images corresponding to respective tilt angles. As a result, as described above, the segmentation module 256 can generate a plurality of series of individualized TEM tilt images, each one in the series corresponding to one respective carbon black aggregate.
[0032] No pre-alignment of aggregate images was required because the segmentation module 256 can identify specific carbon black across a series of TEM tilt images by applying machine vision techniques. Thus, in sharp contrast to existing techniques, fiducial markers are not required in embodiments of the present disclosure. Fiducial markers are opaque objects that can be placed within the field of view of an electron microscope and are generally used to aid alignment in conventional tomography. The TEM image acquisition and processing of the present disclosure can be simplified relative to existing techniques.
[0033] For the selected carbon black aggregate, the segmentation module 256 can also generate a binarized image for each TEM image in the series of personalized TEM tilt images corresponding to the selected carbon black aggregate. Thus, for each TEM tilt image in the series, the segmentation module 256 can convert the grayscale TEM tilt image into a binary TEM image. By way of example, FIG. 4 shows a 0° TEM image (grayscale) and a corresponding 0° binary image. As a result, the segmentation module 256 can generate a series of binarized TEM tilt images for the selected carbon black aggregate. The segmentation module 256 can retain in the memory 278 the imaging data defining the series of binarized TEM tilt images. The segmentation module 256 can generate the binarized images by applying one of a number of techniques for 2D automated image analysis. Specifically, an exemplary technique includes, for each (grayscale) TEM tilt image in the series of personalized TEM tilt images, (i) smoothing the TEM tilt image by applying a median filter having a defined radius (e.g., 3×3 pixels (or radius 1)). The exemplary technique also includes (ii) flattening the resulting TEM tilt image (still in grayscale) by applying a rolling ball algorithm. Flattening can level the background of the image and fill valleys without cutting peaks. Existing 2D image processing techniques for binarizing grayscale images typically lack such background leveling. The exemplary technique further includes (iii) applying a first type of auto-threshold (e.g., IJ_IsoData) to the flattened TEM tilt image (still in grayscale) and applying a second type of auto-threshold (e.g., Triangle) to the flattened TEM tilt image. The application of the first and second types of auto-thresholds results in respective thresholds. Existing 2D image processing techniques for binarizing grayscale images typically provide a simpler auto-threshold approach.(iv) The exemplary technique further includes generating a binary TEM tilt image by applying the respective threshold averages to the flattened TEM tilt image. By applying the respective threshold averages, some gray tones (or pixel values) in the flattened TEM tilt image are assigned pixel values indicative of black, and other gray tones (or pixel values) in the flattened TEM tilt image are assigned pixel values indicative of white. Thus, each pixel value in the flattened TEM tilt image is converted to either black or white, resulting in a binary TEM tilt image. The exemplary technique also includes (v) applying a morphological "open" operation (a combination of an erosion operation followed by a dilation operation) to the binary TEM tilt image. The resulting binary image is configured as a grayscale binary TEM tilt imageTEM tilt image. The thickness of the pixel skin removed in the "open" operation may be configurable. The number of adjacent background pixels required before a pixel is removed from an edge of an object during the erosion operation is also configurable. The number of adjacent foreground pixels required before a pixel is added to an edge of an object during the dilation operation is also configurable.
[0034] The process flow 100 (FIG. 1) also includes a 3D reconstruction stage 130 that can generate a 3D aggregate model of the carbon black aggregate. The modeling system 245 can implement the 3D reconstruction stage 130 via a computing device 250 (FIG. 2). The computing device 250 can include a reconstruction module 260 that includes a selector component 262 that can identify a particular carbon black aggregate to be reconstructed based on a TEM 2D image of the carbon chunk. The selector component 262 can then use the TEM 2D image to determine a number of spheres N for the 3D reconstruction of the particular carbon black aggregate. Each sphere represents a carbonaceous particle that is part of the particular carbon black aggregate. The TEM 2D image can be a no-tilt (0°) TEM tilt image in a series of TEM tilt images of the carbon chunk. The number of spheres N can be defined as a 2D-based estimate of the number of particles in the aggregate (NPA). N can therefore be determined based on the volume (V) of the aggregate.agg ) to the particle volume (V part ) can be determined as the quotient of N=V agg / V part Here, V agg is the aggregate volume estimator (8 / 3)A 2 / P, which is an approximation of the area-perimeter relationship derived for random flocs, where A denotes the area of the aggregate and P denotes the perimeter of the aggregate. part can be determined from the particle size average chord model. More specifically, particle size can be represented by the diameter d of a spherical particle, where d is estimated from A and P as d = απ(A / P). The parameter α is 13.092(P 2 / A) -0.92 or 0.4, whichever is greater. Then, V part =πd 3 Thus, to determine N, the selector component 262 applies 2D analysis to the TEM 2D images to determine V for a particular carbon aggregate. agg and V part It is possible to determine the following.
[0035] After N is determined, the reconstruction module 260 can generate a 3D aggregate model of the particular carbon black aggregate by determining a solution to an optimization problem for an objective function based on the sizes of the N spheres and the positions of the N spheres within a defined volume in three-dimensional space. Determining such a solution may be accomplished via an iterative process that results in model data defining the 3D aggregate model in response to converging to or satisfying another type of termination criterion. Upon convergence, the iterative process also results in a converged value of the objective function. Each iteration in the process may be referred to as a reconstruction step.
[0036] The update component 264 can update the arrangement of the N spheres within the defined volume. The arrangement is defined by a configuration of positions and a configuration of sizes of the spheres within the defined volume. Thus, updating the arrangement can include updating the configuration of positions and updating the configuration of sizes. In other words, updating the arrangement can include adjusting the size and position of each of the N spheres. In some cases, the update approach can include updating a single sphere in each reconstruction step and cycling through the spheres in order when the reconstruction step is performed. Specifically, in iteration j (or the j-th reconstruction step), a single sphere Σ of the N spheres is updated. λ (where λ is an index equal to one of 1, 2...N-1, or N) may be updated while the positions and sizes of the other N-1 spheres remain unchanged. Then, at iteration j+1, a single sphere Σ κ+1 The position of the sphere can be updated, while the positions and sizes of the other N-1 spheres remain unchanged in the next iteration. As the iterations continue, each of the spheres is updated in iteration j+N+1 to the position of the sphere Σ λ may have been individually updated such that ρ is updated again, and the next iteration is implemented, continuing to traverse the set of spheres as the other spheres' positions are updated individually. The update of the sphere's position can be implemented in multiple ways. As an example, the position of the sphere to be updated can be represented by a position vector r that can be defined in a Cartesian coordinate system, such that r can be a triple (x,y,z). Updating the position of that sphere can include modifying x to x+Δx, y to y+Δy, and z to z+Δz, where Δρ=λD(ρ=x,y,z), where D is a quantile and λ is a uniform random number in the interval [0,1]. In some cases, the quantile D=α <d>where α is a parameter that exponentially decays from a maximum to a minimum value as iteration j increases, <d>is the average diameter of the N spheres in the previous iteration. Since α decays as the iterations progress, larger displacements (and subsequent coarser position updates) are implemented earlier in the iteration process, and smaller displacements (and subsequent finer position updates) are implemented later in the iteration process to determine a solution to the optimization problem. Such earlier coarser position updates followed by smaller displacements (or finer position updates) can allow for mitigating or avoiding local minima in the objective function. Additionally, based on iteration j, the size of the single sphere can also be updated. For example, the size of the sphere can begin to be modified after a defined number of reconstruction steps have been implemented. The magnitude of the modification to the diameter of the sphere can decay exponentially as iteration j increases. Other approaches to updating the placement of the spheres can be implemented. For example, updating the placement can include performing a simulated annealing update, performing a genetic algorithm update, or performing a conjugate gradient steepest descent update.
[0037] The selector component 262 can obtain a binarized TEM image of a particular carbon black aggregate. The binarized TEM image can be retained in the memory 278 and corresponds to a binarized TEM tilt image observed at a defined tilt angle. The observed binarized TEM tilt image can be one of a series of individualized binarized TEM tilt images spanning a tilt angle range [-Θ,Θ] at successive tilt angle intervals Δθ. In one example, Θ=70° and Δθ=10°, and the defined tilt angles relate to the set {-70°, -60°, -50°, -40°, -30°, -20°, -10°, 0°, 10°, 20°, 30°, 40°, 50°, 60°, 70°}.
[0038] The selector component 262 can pass the binarized TEM image to an evaluation component 266 included in the reconstruction module 260. The evaluation component 266 can then determine a projection of the arrangement of N spheres onto a plane corresponding to the defined tilt angle. In addition, the evaluation component 266 can also determine a goodness of fit metric based on the binarized TEM image and the projection. The goodness of fit metric can be the number (or, in some cases, a function of the number) of mismatched pixels resulting from pixel-by-pixel differences between the binarized TEM image and the projection. That is, the number of carbon black and background pixels that differ in the projection, pixel-by-pixel, relative to the carbon black and background pixels in the binarized TEM image. FIG. 4 shows an example of pixel-by-pixel differences for a 0° tilt TEM image.
[0039] The selector component 262 can traverse the series of individualized binned TEM tilt images and pass each of the binned TEM tilt images to the evaluation component 266 for determination of a goodness of fit metric at the defined tilt angle of the binned TEM tilt image received by the evaluation component 266.
[0040] After the set of personalized binarized TEM tilt images has been traversed, the evaluation component 266 can update the objective function based on an average of the goodness of fit metric over the set of personalized binarized TEM tilt images. In some cases, as the 3D reconstruction progresses, the evaluation component 266 can determine that the value of the updated objective function is insufficient. In response, the evaluation component 266 can instruct the update component 264 to update the current arrangement of the N spheres. In other words, the evaluation component 266 can cause another reconstruction step to be implemented that includes evaluation of an updated projection based on the updated arrangement. In some cases, also in response to the updated objective function being insufficient, the evaluation component 266 can instruct the update component 264 to reject the current arrangement of the N spheres and return to the previous arrangement of the N spheres before the next update to the arrangement of the N spheres is implemented. By rejecting a current arrangement of N spheres in scenarios where the value of the objective function for such arrangement is insufficient, the reconstruction module 260 may mitigate, or even completely avoid, searching paths on the hypersurface of the objective function that do not lead to an optimization (e.g., minimization) of the objective function. As a result, such rejection of an arrangement(s) of N spheres may result in a more efficient utilization of computing resources, such as processing time (or number of processor cycles) and / or memory storage.
[0041] After implementation of some reconstruction steps, the evaluation component 266 can determine that the value of the objective function is satisfactory. To that end, the evaluation component 266 can determine that the value of the objective function satisfies a convergence criterion. The convergence criterion can specify that the magnitude of the difference between the current value of the objective function and the previous value in the immediately preceding determination of the objective function is equal to or less than a threshold value. In response to determining that the value of the objective function is satisfactory, i.e., that the iterative 3D reconstruction process has converged, the evaluation component 266 can terminate the 3D reconstruction and configure the current model (obtained after the last update to the arrangement of N spheres) as the 3D aggregate model of the particular carbon black aggregate. The evaluation component 266 can optionally terminate the iterative 3D reconstruction process even in the absence of convergence. To that end, the evaluation component 266 can optionally determine whether a termination criterion is satisfied. The termination criterion can, for example, be equal to or less than a threshold number of performed updates N max In some cases, the threshold number N max can be defined as a multiple of the number of spheres (N) in the sphere arrangement, i.e., N max = qN, where q is a natural number (e.g., 40, 50, or 60). max After the updates (or iterations), the changes to the placement of the spheres may be finalized and the iterative 3D reconstruction process may terminate. In one example, when q=50 and N=60, the iterative 3D reconstruction process may thus terminate upon completing 3000 reconstruction steps. By introducing a termination criterion, the amount of computing resources may be limited, thus avoiding inefficient use of those resources (e.g., processor time and / or memory usage).
[0042] The 3D aggregate model of carbon black aggregates generated by applying the iterative reconstruction approach described herein includes the three-dimensional vector position and size (e.g., diameter or radius) of each constituent sphere in the 3D aggregate model. That is, each sphere Σ κ For (κ=1,2.N-1,N), the 3D aggregate model is a 4-tuple (r κ ;σ κ ), where r κ is a position vector in 3D space, and σ κ is a sphere Σ κ The parameter indicates the size of the sphere Σ κ The vector position can be defined in a Cartesian coordinate system. For example, the position vector can be defined in a Cartesian coordinate system, so that r κ is a triple (x κ , y κ , z κ ). The reconstruction module 160 can retain such data defining the 3D aggregate model in memory 278.
[0043] Such a compact form of data defining the 3D aggregate model provides an improvement of the disclosed 3D reconstruction approach over conventional tomography. Indeed, for a typical carbon black aggregate, four parameters per sphere results in on the order of 10-1000 parameters to record surface voxel coordinates, as opposed to 10,000 or more parameters to record surface voxel coordinates derived from conventional reconstructions for sphere-based 3D aggregate reconstructions. Thus, embodiments of the present disclosure utilize computing resources, such as memory storage, more efficiently than existing 3D reconstruction techniques.
[0044] Although the iterative reconstruction approach described above relies on the same series of binarized TEM tilt images for each reconstruction step, the disclosure is not limited in that respect. Indeed, in some embodiments, after a defined number of reconstruction steps have been implemented, the current series of binarized TEM tilt images may be updated to create another series of binarized TEM tilt images. After such updating, subsequent reconstruction steps may be implemented using other series of binarized TEM tilt images until the defined number of reconstruction steps are again implemented. At this point in the iterative reconstruction approach, the other series of binarized TEM tilt images may be updated to generate the next series of binarized TEM tilt images. Subsequent reconstruction steps may be implemented using the next series of binarized TEM tilt images. These successive updates may continue until the iterative reconstruction approach is terminated.
[0045] More specifically, in some embodiments, the reconstruction module 260 and the segmentation module 256 can update the current set of binarized TEM tilt images of the carbon black aggregate. The reconstruction module 260 can determine, for example, via the evaluation component 266, whether the current set of binarized TEM tilt images should be updated. The current set of binarized TEM tilt images corresponds to a set of TEM tilt images observed for the carbon black aggregate. To that end, the reconstruction module 260 can determine whether an update criterion is met. The update criterion can be that a defined number of reconstruction steps have been performed. Other update criteria can also be contemplated. In some examples, the reconstruction module 260 can determine that the update criterion is met, such that the reconstruction module 260 can obtain a binarized TEM image of the carbon black aggregate at a defined tilt angle. The binarized TEM image is part of the current set of binarized TEM tilt images and thus corresponds to the defined tilt angle. The reconstruction module 260 can also determine or otherwise generate a projection of an arrangement of N spheres representing the current 3D model of the carbon black aggregate. Such an arrangement can be determined by the projection of each sphere Σ κ The current set of 4-tuples for (κ=1,2.N-1,N) {(r κ ;σ κ )}. The current 4-tuple does not represent an optimized or otherwise converged 3D model of carbon black aggregates.
[0046] In some embodiments, the reconstruction module 260, via the update component 264, can apply a displacement, for example, to the binarized TEM image. In other embodiments, the reconstruction module 260 can instruct the segmentation module 256 to apply such a displacement. Regardless of the module that applies the displacement, the displacement can be represented by a displacement vector d in terms of a set of multiple displacement vectors. A definition of the set of multiple displacement vectors can be held, for example, in the memory 278. In some cases, each displacement vector of the set of multiple displacement vectors can be a vector that is smaller than or equal to (2N L +1) × (2N L +1) pixels, where L is a natural number. In one example, N L = 5, and 121 displacement vectors form a set of displacement vectors. Such a displacement vector d can be defined in a Cartesian coordinate system with the origin at the geometric center of a square lattice. Thus, the displacement vector is a pair (n x , n y ), where n x and n y are in the interval [-N L , N L ]. The geometric center is located at a pixel, which may be referred to simply as the center pixel for nomenclature reasons. Thus, applying the displacement may include displacing the binarized TEM image by a displacement vector representing the displacement. Applying the displacement results in a displaced binarized TEM image. If the segmentation module 256 applies the displacement, the reconstruction module 260 may receive the imaging data that defines the displaced binarized TEM image. Regardless of how the displacement is applied, the reconstruction module 260 may retain the resulting displaced binarized TEM image in memory 278, for example, in a file system therein.
[0047] The reconstruction module 260, via the evaluation component 266, can determine a goodness of fit metric, for example, based on the displaced binarized TEM image and the projection. As described herein, the goodness of fit metric can be the number (or, in some cases, a function of the number) of mismatched pixels resulting from pixel-by-pixel differences between the binarized TEM image and the projection. Thus, determining the goodness of fit metric can include determining the number of mismatched pixels. The reconstruction module 260 can update a data structure to add the determined goodness of fit metric. The data structure can be retained in memory 278.
[0048] The reconstruction module 260 may continue to traverse the set of the multiple displacement vectors to determine a goodness of fit metric for each displacement applied to the binarized TEM image of the carbon black aggregate at the defined tilt angle. After the goodness of fit metric is determined, the reconstruction module 260 may update a data structure maintained in the memory 278 to add the determined goodness of fit metric. After the set of the multiple displacement vectors is fully traversed, the reconstruction module 260 may identify or otherwise determine a displaced binarized image that corresponds to a satisfactory goodness of fit metric. To that end, the reconstruction module 260 may use a data structure to determine a satisfactory metric and then identify and / or select a displaced binarized image from the memory 278. Since each goodness of fit metric determined during the traversal of the set of the multiple displacement vectors indicates a number of mismatched pixels, a satisfactory goodness of fit metric may indicate a minimum number of mismatched pixels. That is, the displaced binarized image identified may have a minimum pixel mismatch with respect to the projection.
[0049] The reconstruction module 260, via the update component 264, can, for example, update the next series of binarized TEM tilt images to incorporate the identified displaced binarized image. Updating the next series of binarized TEM tilt images can include generating such a series. Once other TEM tilt images (binarized or otherwise) are identified, the reconstruction module 260 can update the next series of binarized TEM tilt images in the memory 278. Thus, the next series of binarized TEM tilt images can be formed by displaced binarized images that each minimize pixel mismatch with a projection at a defined angle of the current 3D aggregate model of the carbon black aggregate.
[0050] The reconstruction module 260 can continue traversing the current series of binarized TEM tilt images and apply the set of multiple displacement vectors to each image in the current series of binarized TEM tilt images. After the current series has been traversed, the reconstruction module 260 can compose the next series of binarized TEM tilt images as the current series of binarized TEM tilt images. Additional reconstruction steps can then be implemented as described herein.
[0051] To demonstrate the validity of the disclosed 3D reconstruction approach, the reconstruction module 260 can generate a 3D aggregate model of a 3D printed assembly of 62 overlapping spheres. Such an assembly represents the structure of carbon aggregates on a macroscopic scale. The 3D aggregate model in such a case may therefore be referred to as a 3D reconstructed model. FIG. 5 shows a photographic projection of the 3D printed assembly mounted on a rotating shaft. Five other photographic projections of the 3D printed assembly are determined and used to create a series of binary images. FIG. 6 shows a 3D reconstructed model view corresponding to the photographic projections shown in FIG. 5. In FIG. 6, the grey shades of the spheres indicate a range of diameters, which are defined in terms of multiples of the average particle size d. Thus, spheres with diameters within a certain range may be depicted as having slightly different sizes but the same grey shade. Darker shades indicate a smaller range of diameters. The average objective function value of the obtained 3D reconstructions was 18%. Without intending to be bound by interpretation and / or speculation, such values may result, in part, from inaccuracies in the tilt angles, slight xy plane rotational misalignments, and possible projection distortions.
[0052] Further, the segmentation module 256 can generate simulated binary images for validation of the 3D reconstruction approach of the present disclosure based on the sphere data output from the 3D printed assembly. In one example, a series of 15 simulated binary tilt images obtained for a [-70°, 70°] tilt angle range and Δθ=10° are provided to the 3D reconstruction approach. The average objective function value at convergence (or otherwise termination) of the 3D reconstruction approach for five separate trial reconstructions ranged from 2.3% to 3.1%. The trial reconstructions are 3D reconstructions implemented according to aspects described herein. Thus, the average objective function value is obtained by averaging the converged (or otherwise terminated) objective function values for each 3D reconstruction. Such value of the average objective function establishes the validity of the 3D reconstruction approach based on optimizing the size and position of the sphere within a defined volume as described herein.
[0053] To further validate the 3D reconstruction approach of the present disclosure, the reconstruction module 260 can reconstruct the modeled carbon black aggregate using a conventional TEM tomography reconstruction. FIG. 7A is a no-tilt (0°) TEM micrograph of the carbon black aggregate. FIG. 7B is a no-tilt (0°) tomogram of the carbon black aggregate determined using a conventional TEM tomography reconstruction. Such conventional reconstruction relies on a voxel-based approach and limited segmentation time. FIG. 7C is a no-tilt (0°) sphere-based 3D reconstruction of the carbon aggregate according to an embodiment of the present disclosure. As discussed above in connection with FIG. 6, the gray shaded tones of the sphere indicate the diameter range. From FIG. 7B, it is readily apparent that the conventional voxel-based reconstruction cannot achieve good surface definition with limited segmentation time. There is good agreement in the macroscopic properties between the conventional tomography reconstruction and the sphere-based 3D reconstruction. Here, good agreement of macroscopic properties refers to good agreement of the spatial extent of the aggregates in a first in-plane direction (e.g., x direction) and a second in-plane direction (e.g., y direction) on the imaging plane, and also to good agreement of the curvature of the aggregate boundaries (in the continuous case, and in the coarse "voxelized" voxel-based reconstruction case).
[0054] As described herein, a series of TEM tilt images acquired through a tilt angle range [-Θ, Θ] at a particular Δθ interval can be used for 3D reconstruction of carbon black. Thus, the reconstruction module 260 can use various series of TEM tilt images according to different tilt angle intervals and tilt angle ranges to reconstruct the 30 N330 carbon black aggregates marked in FIG. 3. As described above, the 3D reconstruction process is iterative, potentially converging and generating model data that defines the 3D aggregate model. Upon convergence, the convergence value of the objective function involved in the 3D reconstruction process is also determined. FIG. 8 shows a scatter plot 800 of converged objective function values averaged over the 30 N330 carbon black aggregates, respectively, using a series of TEM tilt images acquired at five tilt angle intervals (2°, 4°, 8°, 16°, and 32°) in the tilt angle range [-70°, 70°]. FIG. 8 also shows a scatter plot 850 of converged objective function values averaged over 30 N330 carbon black aggregates using a series of TEM tilt images acquired at four tilt angle ranges ([-40°, 40°], [-50°, 50°], [-60°, 60°], [-70°, 70°]) with tilt angle interval Δθ=2°, respectively. The average objective function results show that 3D reconstruction using TEM tilt images acquired at tilt angle interval Δθ=8° performs nearly as well as 3D reconstruction using TEM tilt images acquired at tilt angle interval Δθ=2°. Thus, with a quarter of the number of images acquired, the Δθ=8° series can result in a substantial reduction in data acquisition time compared to the Δθ=2° series. The reduction in the number of tilt TEM images achieved by reducing the tilt angle range does not result in comparable performance. Thus, image acquisition over the tilt angle range [-70°, 70°] may be desirable. Based on the results shown in FIG. 8, a 3D reconstruction can be based on a series of TEM tilt images acquired using the tilt angle range [−70°, 70°] and Δθ=10°.
[0055] FIG. 9 shows seven reconstructed carbon black aggregate samples from the selected 30 N330 carbon black aggregates marked in FIG. 3. Again, the gray shades of the spheres indicate the diameter range. The structural complexity of the seven aggregates increases from left to right (in the readout direction) across FIG. 9. There is good agreement between the 3D model projections and the 2D images, with good agreement in pixel coverage between the TEM image at 0° and the 3D model projection at 0°. Furthermore, it is observed that the aggregates with lower structure appear to have somewhat too few particles to adequately model the volume and surface of the carbon black aggregate, while the aggregates with higher structure have an adequate number of particles. Without intending to be bound by interpretation or inference, such observations may arise from using an estimate of the number of particles based on existing 2D analysis, as described herein.
[0056] The 3D reconstruction approach of the present disclosure can be implemented for multiple separate series of TEM tilt images of carbon black (e.g., N330 carbon black) for many individual carbon black aggregates depicted in the series. A first of the multiple separate series is the series including the non-tilted TEM micrographs shown in FIG. 3. In one exemplary scenario, the reconstruction module 260 can apply the 3D reconstruction process of the present disclosure to a total of 265 potentially reconstructable aggregates, where potentially reconstructable aggregates refer to carbon black aggregates that are visible in all projections across the series of TEM tilt images, for which the segmentation module 256 can determine aggregate binaries that have a satisfactory match of the template in each image in the series of TEM tilt images. Again, the 3D reconstruction process is iterative and, in some cases, converges, resulting in a converged value of the objective function involved in the 3D reconstruction process. The resulting distribution of converged average objective functions is shown in FIG. 10. By defining a quality threshold (e.g., 25%, 26%, 27%, or 30%), it can be determined whether the reconstruction is satisfactory or unsatisfactory. Satisfactory reconstructions have converged average objective functions below the quality threshold, while poor reconstructions have converged average objective functions above the quality threshold. As shown in FIG. 10, 256 reconstructions are satisfactory. Each of the satisfactory reconstructions is deemed of sufficient quality to contribute to the statistical analysis, has a final average objective function value of less than 26%, and contributes to the ensemble converged average objective function value of 10.8%.
[0057] Computing device 250 (FIG. 2) can include one or more processors (not shown in FIG. 2), which can be assembled into a single core or multiple cores. As an illustration of the computational efficiency of the 3D reconstruction approach of the present disclosure, computing device 250 can perform 3D reconstruction of 256 carbon black aggregates that resulted in a satisfactory reconstruction in 48 hours of single-core processor computation time. The 3D reconstruction of the relatively small number of large carbon black aggregates consumed the majority of the computation time. FIG. 11 shows the distribution of 3D reconstruction computation times.
[0058] For each satisfactory reconstruction, the computing device 250 can include an analysis module 270 that can use the determined 3D agglomerate model to determine the 3DSSA for the carbon black agglomerate corresponding to the satisfactory reconstruction. Here, for purposes of illustration, the 3D specific surface area (3DSSA) of a carbon black agglomerate refers to the total surface per unit mass, where the total surface accounts for the surface adjacent to the voids within the carbon black agglomerate. The 3DSSA distribution for the 256 satisfactory reconstructions is shown in FIG. 12. The 3DSSA distribution is 2 / g resulting in a volume weighted 3DSSA.
[0059] The post-reconstruction analysis is not limited to determining the 3DSSA for the reconstructed carbon black aggregate. Indeed, after the 3D aggregate reconstruction is implemented, a general analysis of the morphology of the reconstructed carbon black aggregate can be implemented in the post-reconstruction analysis stage 140 of process flow 100 (FIG. 1). More specifically, in the post-reconstruction analysis stage 140, various morphological properties of the reconstructed carbon black aggregate can be determined using the 3D aggregate model determined in the 3D reconstruction of the carbon black aggregate. As described above, the 3D aggregate model includes the three-dimensional vector position and size (e.g., diameter or radius) of each constituent sphere within the 3D aggregate model.
[0060] The modeling system 245 may also implement the post-reconstruction analysis stage 140 via the computing device 250 (FIG. 2). To that extent, the analysis module 270 may determine various structural properties of the reconstructed carbon black aggregates, which may be determined using the 3D aggregate model. Examples of structural properties include 3DSSA (as described above), aggregate volume, aggregate convex hull volume, void volume, 3D VV, aggregate extent in X (X extent ), the aggregate extent at Y (Y extent ), and the aggregate extent in Z (Z extent ), and "relative Z" (Z rel ) where Z rel is the anisotropy metric defined as the z dimension of the reconstituted carbon black aggregate relative to the average of the x and y dimensions of the aggregate, i.e., Z rel =2Z extent / (X extent +Y extent ).
[0061] The analysis module 270 can determine the mass and specific surface area distribution of overlapping sphere volumes of the reconstructed carbon black aggregates through 3D gridding (e.g., voxelization). The analysis module 270 can determine the 3DSSA of the overlapping spheres by identifying and counting the aggregate surface voxels and correcting the projected surface area coverage of the voxels at different angles (the correction factor can be averaged over spheres of variable voxel radius). More specifically, the analysis module 270 can apply a conversion (or correction) factor γ to some aggregate surface voxels to determine the surface area of the overlapping spheres. The surface area can then be used in determining the 3DSSA. The conversion factor γ represents an empirical relationship between the number of surface voxels and the surface area of a spherical particle. Thus, γ can be empirically determined by converting the surface voxels to the surface area of a sphere having a range of diameters spanning an interval that includes the average diameter of the particles in a type of carbon black aggregate under analysis (e.g., N330 carbon black aggregates). In this way, a conversion factor γ=1.30 can be obtained and used in this disclosure. The analysis module 270 can determine the occluded volume by determining an approximation to the aggregate convex hull. An analytical expression for the aggregate convex hull is not available for volumes enclosed by discrete spheres. Thus, without intending to be constrained by the modeling, the convex hull can be determined by removing masses outside the bounding plane in rotational orientations varying by a defined angle increment (e.g., 6°) around all axes in the Cartesian coordinate system, summing the remaining volumes as the convex hull volume, and subtracting the aggregate volume to arrive at the intra-convex hull boundary set void volume. The intra-aggregate void volume divided by the aggregate volume defines the 3DVV.
[0062] Merely by way of example, analysis module 270 may determine aggregate volume, void volume, and 3DSSA according to the embodiments described herein for the 3D reconstruction of the 30 N330 carbon black aggregates marked in FIG. 3. Reconstruction module 260 generated the 3D reconstruction using TEM tilt images acquired using [±70°, 70°] and Δθ=10°. Analysis module 270 determined the aggregate size range in X, Y, and Z. Data for the 30 reconstructed aggregates is shown in Table 1. In that table, "Agg.ID" refers to the number adjacent to the marked aggregate in FIG. 3. [Table 1]
[0063] According to existing analyses of carbon black aggregates available in the existing carbon black literature, the degree of anisotropy of carbon black aggregates increases with aggregate structure. Figure 13 shows the Z-axis distribution for the 30 N330 carbon black aggregates identified in Table I and marked in Figure 3. rel It is observed that all aggregates have less degree of anisotropy in Z than in X and Y. There is also some dependence of anisotropy on aggregate structure as measured via 3DVV, with more highly structured aggregates tending to have more anisotropy (i.e., lower Z rel Thus, the results obtained using the disclosed 3D reconstruction approach and shown in FIG. 13 support existing findings and provide a direct measure of aggregate anisotropy. That is, anisotropy is a function of the Z rel and that metric can be determined directly from 3D reconstructions according to aspects of the present disclosure. Such direct measurements are not available in existing analyses of carbon black aggregates because they are based on 2D data and only infer anisotropy in the third dimension (e.g., z-direction).
[0064] For 3D aggregate models according to embodiments of the present disclosure, aggregate volume ("Agg.Vol" in Table 1) can be calculated directly. As shown in FIG. 14, particularly in the inset, 3D aggregate volume correlates with 2D-based volume for aggregates of lower volume (and generally lower structure), with the 2D-based volume increasing relative to the 3D-based volume as volume increases. Without intending to be bound by theory or modeling, this can be understood by considering the ramifications of anisometry on 2D-based volume measurements. Specifically, existing 2D estimators consider aggregate isometry, such that a reduction in aggregate extent in the Z direction for the case of isometric aggregates makes the 2D-based calculation larger than the 3D-based volume determined directly by the 3D reconstruction of the present disclosure.
[0065] If the density is known, the 3DSSA can be determined from a voxelization of the 3D aggregate model (as described herein) and can be directly related to bulk colloid measurements. Individual aggregate 3DSSA values shown in Table 1 are based on 30 m 2 / g~85m 2 As shown in FIG. 15, the distribution of 3DSSA values for the 30 reconstituted N330 carbon black aggregates is approximately normal or log-normal. The volume weighted average of the 3DSSA is 52.5 m 2 / g. This value is for N330 carbon black (75m 2 30% smaller than a typical statistical thickness surface area (STSA) value of about 55m / g. Without intending to be bound by interpretation, such difference may be due to the interior regions of the reconstituted carbon black aggregates having significant excess mass and significantly larger diameter spherical particles. Such factors, individually and in combination, reduce the modeled specific surface area (SSA). Note that analysis module 270 can determine the SSA by counting boundary pixels just inside the carbon aggregate. If instead, spatial boundary voxels just outside the carbon aggregate are counted, the 3DSSA value is about 55m. 2 / g, which is still less than the typical STSA value.
[0066] The existing 2DV' / V supports the expected fundamental relationship between OAN and relative void volume, which is the aggregate void volume (denoted as V') divided by the aggregate volume (denoted as V). For this relationship, the commonly accepted approximation for carbon black density of 1.9 g / cm 3 You can convert volume to mass using the OAN units (cm 3 When adjusted to a 2DV' / V ratio (1 / 100g), the theoretical slope of OAN to true relative void volume is 100 / 1.9 ≈ 53. 2DV' / V can be referred to as the aggregate extinction coefficient or structure factor and can represent the relative void volume. Such a relationship is derived from a 2D-based approximation of the diameter of the space occupied by the aggregate. Without intending to be bound by speculation, it is expected that 3DVV will follow a similar relationship if it is a useful measure of relative void volume. In that regard, FIG. 16 indeed shows a linear relationship for the 30 aggregates measured in this disclosure. The slope of the linear relationship is close to 1 with a small offset. The 3DVV values are approximately 0.3 units lower than the 2DV' / V values (calculated from a single 0° aggregate projection). The weighted void volume determined via convex hull approximation was 1.39, 18% less than the 2DV' / V relative void volume of 1.69 for these 30 N330 carbon black aggregates.
[0067] Without intending to be bound by interpretation or speculation, the combination of a 25%-30% lower specific surface area and an 18% lower 3DVV compared to expectations from the 2D analysis may be due to the inclusion of excess mass near the center of the 3D aggregate model. Filling of small spaces not visible in the binary due to overlap of mass of neighboring aggregates may slightly increase mass and decrease surface area. The combined effect of increased mass and decreased surface area may decrease the determined 3DSSA and 3DVV.
[0068] Utilizing the morphological properties of carbon black allows for the analysis of how these properties affect the carbon black-rubber interaction through occlusion. Occluded volume can be approximated as the volume contained within a convex hull surrounding the carbon black aggregate. Existing carbon black literature contains relationships that directly link carbon black occluded volume measurements to oil absorption (also referred to as DBP). As mentioned, OAN is a bulk measure of the amount of oil required to just fill the intra-aggregate voids associated with the carbon black aggregate. The existing literature also raised issues regarding the concept of rubber occluded "within" the carbon black aggregate (and shielded from strain to a certain geometric degree) since each intra-aggregate rubber volume element may actually experience some degree of occlusion measured on a continuum depending on the stress shielding at that location.
[0069] The question of what volume is "in" the aggregate can be addressed with data showing the 3D mass distribution of the carbon black aggregates. Such distribution can be modeled based on the spatial distribution of the spheres forming the 3D aggregate model according to the present disclosure. Analysis module 270 (FIG. 2) using such data can implement nanoscale studies including finite element analysis (FEA) of rubber deformation in proximity to the 3D aggregate model of the carbon black aggregate according to the present disclosure. Analysis of many other applications can also be implemented using such data. Carbon black isolated from rubber compounds can also be analyzed, since carbon black aggregates degrade to some extent during rubber compounding. Thus, the data defining the 3D aggregate model described herein provides a rich basis of 3D carbon black aggregate morphological data that can be readily applied to model reinforcement in rubber and also to guide product development efforts.
[0070] In addition to occlusion and reinforcement, many other rubber properties are substantially dependent on carbon black morphology. Again, having access to morphology data indicative of the morphological properties of carbon black aggregates and model data representing 3D carbon black mass distribution can elucidate aspects of properties such as permeability, conductivity, and hysteresis, and can allow the effects of one or more of those properties to be predicted and / or engineered. By way of example only, existing literature on carbon black has investigated the effects of filler morphology on permeability via simulated aggregate modeling and diffusion. Data indicative of 3D carbon black morphology can validate diffusion models in that existing literature and can also elucidate geometric optimization of carbonaceous solid barriers to gas permeation. Data indicative of aggregate mass distribution in 3D, particularly around carbon black aggregates, can allow the filler-filler network properties of the polymer to be established. Such networking properties can affect static properties such as conductivity, and can also affect hysteresis transient filler networking (known as the Payne effect). The wealth of morphology data provided by the 3D reconstruction approach of the present disclosure is not available from existing tomographic reconstruction techniques. Thus, embodiments of the present disclosure can provide a powerful data framework applicable to understanding and engineering such phenomena. Furthermore, as described herein, such a powerful data framework can be used to identify the carbon black type of carbon black aggregates based on machine learning predictive models. Such models can be trained using the wealth of data provided by the data framework. Thus, embodiments of the present disclosure can provide a variety of practical analytical tools for the study of carbon black aggregates.
[0071] With further reference to FIG. 2, the computing device 250 may also include a reporting module 274 that may provide a 3D aggregate model and / or morphological data of the carbon black aggregates. Such information may be provided in a number of ways. In some embodiments, to provide the 3D aggregate model, the reporting module 274 may retain the 3D aggregate model in one or more memory devices (e.g., memory 278) and may constitute an API for accessing the 3D aggregate model. Additionally, or in some cases, the reporting module 274 may provide the 3D aggregate model by causing the presentation of a visualization of the 3D aggregate model in a user interface. The reporting module 274 may cause the display device 280 to present such a user interface. As shown in FIG. 2, the display device 280 may be operatively coupled to the computing device 250. Such coupling may be provided by a wired interface and / or a wireless interface (such an interface is represented by a segment with an open arrow). In other cases, the display device 280 may be integrated into the computing device 250. Similarly, to provide morphological data, e.g., values of morphological properties of carbon aggregates, the reporting module 274 can maintain the values in one or more memory devices (e.g., memory 278) and the API can be configured to access one or more values of the particular morphological properties. Additionally, or in other embodiments, to provide such values, the reporting module 274 can cause the presentation of a visual element in a user interface that indicates one or more values of the one or more morphological properties. Again, that user interface can be presented on a display device 280. Additionally, or in yet other embodiments, to provide such values, the reporting module 274 can cause the sending of an electronic message (e.g., an email) conveying one or more values of the one or more morphological properties.The reporting module 274 can also cause the presentation of a notification that the 3D aggregate model is available for analysis, including a visualization of the 3D aggregate model.
[0072] The 3D reconstruction of the carbon black aggregate and the post-reconstruction analysis of the carbon black aggregate can be used to identify the type of carbon black of the carbon black aggregate. To that end, as shown in FIG. 17, the post-reconstruction analysis 140 can be extended to include sub-stages that can make it possible to identify the type of carbon black corresponding to the carbon black aggregate based on the value of a morphological property of the carbon black aggregate. In a first sub-stage 1710, a morphological feature can be determined. The morphological feature can be the value of a morphological property P of the carbon black aggregate. Thus, in the first sub-stage 1710, a feature vector f can be determined, which is the sum of the properties P1, P2, ...P M , each component may have a value determined using the 3D aggregate model of the carbon black aggregate determined in stage 130, as described herein. In some cases, M=7 and the morphological property may be extent , Y extent , Z extent , Z rel , aggregate volume, 3DVV, and 3DSSA. The disclosure is not limited to that number of morphological features and fewer or more than seven morphological features may be contemplated. A second sub-stage 1720 may identify the type of carbon black by applying a predictive model to the feature vector f. The predictive model may be a machine learning model configured (e.g., trained) to designate carbon black aggregates as being of a particular type of carbon black. For example, the predictive model may be a classification model configured (e.g., trained) to solve a multi-task classification problem.
[0073] To generate such a predictive model, operating environment 200 (FIG. 2) can be used to generate data sets corresponding to each carbon black aggregate for known types of carbon black (e.g., N110 carbon black, N220 carbon black, N330 carbon black, N550 carbon black, N660 carbon black, N772 carbon black, N880 carbon black, N990 carbon black, etc.). Each data set in the data set can be labeled according to the type of carbon black, and a feature vector f=(P1, P2, ... P M ). Such a labeled dataset may be referred to as labeled data and may be maintained in one or more memory devices (referred to as a labeled data repository). A predictive model may then be trained to solve a multi-task classification problem on the feature vectors corresponding to multiple values of the morphological properties of the carbon black aggregates.
[0074] FIG. 18 illustrates an example system 1800 for generating a predictive model configured to designate a carbon black aggregate as being a particular type of carbon black, according to one or more embodiments described herein. As described above, the predictive model may be embodied in a classification model configured to solve a multi-task classification problem. The system 1800 may include a training module 1810. The training module 1810 may implement supervised, unsupervised, and / or semi-supervised (e.g., reinforcement-based) machine learning models. The training module 1810 may obtain different types of training data, for example, via a harvesting component 1814. Because the predictive model may be trained to solve a multi-task classification problem for feature vectors corresponding to multiple values of morphological properties of the carbon black aggregates, the harvesting component 1814 may obtain labeled data 1812 from one or more memory devices 1804 (referred to as a labeled data repository 1804). The labeled data 1812 may define a plurality of labeled feature vectors from a group of a plurality of labeled feature vectors defined by the labeled data 1808. The feature vector may be labeled, for example, as part of a 3D reconstruction of a carbon black aggregate of a known type of carbon black (e.g., N110 carbon black, N220 carbon black, N330 carbon black, N550 carbon black, N660 carbon black, N772 carbon black, N880 carbon black, N990 carbon black, or the like). The labels of the labeled feature vector may include text elements. Each label of the labeled feature vector specifies a type of carbon black, for example, N110 carbon black, N220 carbon black, N330 carbon black, N550 carbon black, N660 carbon black, N772 carbon black, N880 carbon black, N990 carbon black, or the like.
[0075] The training module 1810 also includes a constructor component 1816 that can operate on the data 1812 acquired by the harvesting component 1814. By operating on the data 1812, the constructor component 1816 can train a detection model using at least a subset of the labeled images 1808. As described above, the predictive model can be a classification model that can be trained to solve a multi-task classification problem and designate carbon black aggregates as belonging to one of a number of categories, each category representing a type of carbon black. To train the predictive model, the constructor component 1816 can use the data 1812 to determine a solution to an optimization problem with respect to a cost function. The solution can be determined iteratively, where in each iteration, the cost function results in a value based on an evaluation of the difference between the known label for each carbon black aggregate and a predicted label for each carbon black aggregate. The predicted label is predicted or otherwise determined by the current iteration of the classification model, i.e., the classification model defined in that current iteration. After a solution is determined, the solution results in model parameters that minimize the cost function. The model parameters define a trained predictive model 1820. The training module 1810, e.g., via the constructor component 1816, can persist the trained predictive model 1820 in one or more memory devices 1830 (referred to as a model repository 1830).
[0076] In some embodiments, the training module 1810 can be hosted by the computing device 250, and the labeled data repository 1804 and the model repository 1830 can be part of the memory 278. In other embodiments, another computing device or a system of computing devices operatively coupled to the computing device 250 can include the training module 1810, as well as the labeled data repository 1804 and the model repository 1830.
[0077] The trained predictive model (e.g., predictive model 1820) can be used to identify the type of carbon black to which the carbon black aggregate pertains. To that end, the carbon black aggregate can be examined and modeled using electron microscopy according to the embodiments described herein in connection with process flow 1700 (FIG. 17). Accordingly, operating environment 1900 can be utilized to generate a 3D aggregate model of the carbon black aggregate by determining a 3D reconstruction of that carbon black aggregate. The 3D aggregate model can be generated by implementing steps 110 through 130 of process flow 1700 via computing device 250 (FIG. 19). As described herein, analysis module 270 can generate values for each morphological property of the carbon black aggregate based on the 3D aggregate model. Analysis module 270 can cast such values as a feature vector f, where each component of f corresponds to a morphological property P of the carbon black aggregate and has a value equal to the value determined by analysis module 270 for morphological property P. As described herein, f can be a function of the properties P1, P2, ...P. M , where such characteristics may be the same as the morphological characteristics used to generate a training dataset for the predictive model 1920. In FIG. 19 , M tuples defining an instance of the feature vector f are shown as features 1904. That is, a first carbon black aggregate to be studied, modeled, and analyzed according to the process flow 1700 may have a first feature 1904, e.g., f=(P1, P2, ... P M ), a second carbon black aggregate to be investigated, modeled, and analyzed according to process flow 1700 is f=(P1, P2, ... P M As mentioned above, in some cases, M=7 and the morphological characteristics are extent , Y extent , Z extent , Z rel , aggregate volume, 3DVV, and 3DSSA.
[0078] For a particular carbon black aggregate, the analysis module 270 can determine a characteristic 1904 and pass the determined characteristic 1904 to the carbon black (CB) identification module 1910. In some cases, the characteristic 1904 can be passed by sending data indicative of the characteristic 1904. In other cases, the characteristic 1904 can be passed by reference. That is, the analysis module 270 can hold the characteristic 1904 in memory 278 (FIG. 19) and then send to the CB identification module 1910 a datum or data that points to the memory block in which the characteristic 1904 is held in memory 278 (FIG. 19). The CB identification module 1910 can then read or otherwise obtain the characteristic 1904 from that memory block in memory 278.
[0079] The CB identification module 1910 may apply a predictive model 1920 to the feature 1904. In response to applying the predictive model 1920, the CB identification module 1910 may generate a number of numerical weights corresponding to respective types of carbon black. Such respective types and their number may be the same as the group and number of types used to train the predictive model 1920. Based on the number of numerical weights, the CB identification module 1910 may designate the carbon black aggregate associated with the feature 1904 as being of a certain type of carbon black. For example, the type of carbon black may have the largest weight of the number of numerical weights. The CB identification module 1910 may then cause the display device 280 to present a notice or indicia indicating the type of carbon black.
[0080] In some embodiments, rather than analyzing the carbon black aggregates individually, the computing device 250 of FIG. 19 may analyze a group of carbon black aggregates to predict one or more types of carbon black that are representative of the group of carbon black aggregates. To that end, respective sets of features 1904 may be determined for the carbon black aggregates that form the group of carbon black aggregates. For each of the sets, the CB identification module 1910 may determine a respective prediction of the type of carbon black. The CB identification module 1910 may then evaluate the respective predictions, and based on the evaluation, the CB identification module 1910 may determine one or more representative types of carbon black for the group of carbon black aggregates. The CB identification module 1910 may rank the one or more respective types, and may then cause the display device 280 to present a notice or indicia indicative of the ranking of the one or more respective types.
[0081] In view of the aspects described herein, an exemplary method that may be implemented in accordance with the present disclosure may be better understood, for example, with reference to the flowcharts in FIGS. 20-24. For ease of explanation, the exemplary method disclosed herein is presented and described as a series of blocks (each block represents, for example, an action or operation in the method). However, the exemplary method is not limited by the order of the blocks and associated acts or operations, since some blocks may occur in different orders and / or simultaneously with other blocks than those illustrated and described herein. Furthermore, not all illustrated blocks and associated action(s) may be required to implement an exemplary method according to one or more aspects of the present disclosure. Two or more of the exemplary method (and any other method disclosed herein) may be implemented in combination with each other. It should be noted that the exemplary method (and any other method disclosed herein) may alternatively be represented as a series of interrelated states or events, such as in a state diagram.
[0082] FIG. 20 illustrates an example of a method 2000 for providing 3D reconstruction and analysis of carbon black aggregates, according to one or more embodiments of the present disclosure. A computing device or a system of computing devices can implement the example method 2000 in whole or in part. To that end, each of the computing devices includes computing resources capable of implementing at least one of the blocks included in the example method 2000 and other methods described herein. The computing resources include, for example, a central processing unit (CPU), a graphics processing unit (GPU), a tensor processing unit (TPU), memory, disk space, input bandwidth, and / or output bandwidth, interface(s) (such as an I / O interface or API, or both), controller device(s), power supply, combinations of the foregoing, and / or similar resources. A computing device can embody a computing device 250 and thus host at least one of the capture module 252, the segmentation module 256, the reconstruction module 260, the analysis module 270, and the reporting module 274. Thus, a computing device may implement the example method 2000 by executing one or more instances of such modules.
[0083] In block 2010, the computing device may identify carbon black aggregates based on the TEM 2D image of the carbon mass, in some cases, the computing device may apply one or more machine vision techniques, including object localization techniques such as template matching as described herein.
[0084] In block 2020, the computing device calculates a number N of spheres for the 3D reconstruction of the carbon black aggregates. s To that end, according to embodiments described herein, a computing device can determine the volume (V agg ) and particle volume (V part ) can then be determined by the computing device. agg V part By determining the quotient when dividing by N s That is, N s =V agg / V part It is.
[0085] At block 2030, the computing device may generate a 3D agglomerate model of the carbon black agglomerate by determining a solution to an optimization problem for an objective function based on the size of the spheres and the positions of the spheres within the defined volume. In some cases, to generate the 3D agglomerate model in such a manner, the computing device may implement an example method 2100 shown in FIG.
[0086] In block 2040, the computing device can use the 3D aggregate model to determine values for each of one or more morphological properties of the carbon black aggregate. The one or more morphological properties can be determined using an aggregate inequality metric (e.g., Z rel ), aggregate volume, relative 3D void volume (3DVV), or 3D specific surface area (3DSSA).
[0087] At block 2050, the computing device may provide values for each of the one or more morphological characteristics. In some embodiments, providing such values may include retaining the values in one or more memory devices and configuring an API to access one or more values of particular ones of the one or more morphological characteristics. In addition, or in other embodiments, providing such values may include causing a display device to present a visual element indicative of the one or more values of the one or more morphological characteristics. In addition, or in yet other embodiments, providing such values may include sending an electronic message (e.g., an email) communicating the one or more values of the one or more morphological characteristics.
[0088] At block 2060, the computing device can cause the presentation of a notification that the 3D aggregate model is available for analysis. The analysis can include visualization of the 3D aggregate model in a user interface presented on a display device (e.g., display device 280). In some cases, the display device can present the notification in its user interface or another user interface.
[0089] 21 illustrates an example of a method 2100 for iteratively generating a 3D aggregate model of a carbon black aggregate, in accordance with one or more embodiments of the present disclosure. A computing device or a system of computing devices may implement the example method 2100 in whole or in part. A computing device that implements the example method 2000 may also implement the example method 2100. Thus, a computing device may implement the example method 2100 by executing the reconstruction module 260.
[0090] In block 2110, the computing device may update the placement of the defined number of spheres within the defined volume. The defined number of spheres may be determined in block 2020 of the example method 2000 (FIG. 20). s The arrangement may be defined by a configuration of positions and a configuration of sizes of a plurality of spheres within a defined volume. Thus, updating the arrangement may include updating the configuration of positions and updating the configuration of sizes according to aspects described herein. For example, updating the arrangement may include modifying a position of a first one of the spheres and maintaining a second position of a second one of the spheres. Additionally, or possibly (based on the current iteration or reconstruction step), modifying the size of the first sphere and maintaining the second size of the second sphere. In some embodiments, updating such an arrangement may include performing a simulated annealing update, performing a genetic algorithm update, or performing a conjugate gradient steepest descent update.
[0091] In block 2120, the computing device can obtain a binarized TEM image of the carbon black aggregate. The binarized TEM image corresponds to a TEM tilt image observed at a defined tilt angle in a series of TEM tilt images. The observed tilt TEM image can be one of a series of TEM tilt images spanning a tilt angle range [-Θ,Θ] at successive tilt angle intervals Δθ. In one example, Θ=70° and Δθ=10°, and the defined tilt angles relate to the set {-70°, -60°, -50°, -40°, -30°, -20°, -10°, 0°, 10°, 20°, 30°, 40°, 50°, 60°, 70°}.
[0092] At block 2130, the computing device may determine a projection of an arrangement of the defined number of spheres onto a plane corresponding to the defined tilt angle.
[0093] At block 2140, the computing device may determine a goodness of fit metric based on the binarized TEM image and the projection. The goodness of fit metric may be the number (or, in some cases, a function of the number) of mismatched pixels resulting from pixel-by-pixel differences between the binarized TEM image and the projection. Thus, determining the goodness of fit metric may include determining the number of mismatched pixels.
[0094] At block 2150, the computing device may determine whether another observed TEM tilt image is available for processing in the series of TEM tilt images. A positive determination ("yes" branch) may cause the flow of the example method 2100 to be directed to block 2120 for another iteration of the implementation of blocks 2120, 2130, 2140, and 2150. A negative determination ("no" branch) at block 2150 may cause the flow of the example method 2100 to be directed to block 2160, where the computing device may update the objective function based on an average of the goodness of fit metric over the series of TEM tilt images.
[0095] At block 2170, the computing device may determine whether the value of the objective function is satisfactory. To that end, the computing device may determine whether the value of the objective function satisfies a convergence criterion. The convergence criterion may provide that the magnitude of the difference between the value and the previous value in the immediately preceding determination of the objective function is equal to or less than a threshold. A positive determination ("yes" branch) may indicate convergence and cause the flow of the exemplary method 2100 to be directed to block 2190, where the computing system may configure the current model as a 3D aggregate model of carbon black aggregates. A negative determination ("no" branch) at block 2170 may cause the flow of the exemplary method 2100 to be directed to optional block 2180, where the computing device may optionally determine whether a termination criterion is satisfied. As described herein, the termination criterion may be, for example, a defined number N of updates performed. max The defined number N can be expressed as max is, for example, the number of spheres in the volume (N s (sometimes denoted as N), i.e., N max =qN s where q is a natural number (e.g., 40, 50, or 60). Therefore, N max After the update (or iteration), the changes to the arrangement of the defined number of spheres may be terminated. Although not shown in FIG. 21 , in some cases, in response to a negative determination at block 2170, the computing device may (or may have the computing device reject) the updated arrangement of the defined number of spheres. A negative determination ("no" branch) at optional block 2180 causes the flow of the example method 2100 to be directed to block 2110 for the next update to the current arrangement of the defined number of spheres and the subsequent determination of an updated value of the objective function. In a scenario where the current arrangement of the defined number of spheres is rejected (or configured for rejection), the next update may be based on the previous arrangement of the defined number of spheres instead of the current arrangement. A positive determination ("yes" branch) at block option 2180 causes the flow of the example method 2100 to be directed to block 2190, where the computing system may configure the current model as a 3D aggregate model of carbon black aggregates.
[0096] Although the exemplary method 2100 is described as using the same set of binarized TEM tilt images across reconstruction steps, the disclosure is not limited in that respect. As described herein, in some embodiments, after a defined number of reconstruction steps have been implemented, the set of binarized TEM tilt images may be updated to create another set of binarized TEM tilt images. Subsequent reconstruction steps may be implemented using the other set of binarized TEM tilt images until the defined number of reconstruction steps are again implemented. At that point in the iterative process, the other set of binarized TEM tilt images may be updated to generate the next set of binarized TEM tilt images. Such a sequence of updates may continue until the exemplary method 2100 ends. As described herein, with respect to the exemplary method 2100, the reconstruction steps correspond to implementations of blocks 2110-2160.
[0097] More specifically, in some embodiments, as shown in FIG. 21, a negative determination at block 2180 can optionally direct the flow of the exemplary method 2100 to perform the exemplary method 2200 shown in FIG. 22. In other words, in those embodiments, the exemplary method 2200 can be implemented before implementing the next reconstruction step. The exemplary method 2200 is an example of a method for updating a series of binarized TEM tilt images during or as part of an iterative process for generating a 3D aggregate model of carbon black aggregates. A computing device implementing the exemplary method 2100 (FIG. 21) can also implement the exemplary method 2200.
[0098] In block 2205, the computing device may determine whether the current series of binarized TEM tilt images of the carbon black aggregate should be updated. As shown in FIG. 22 and described herein, in some cases, block 2205 may be implemented in response to a "no" branch from optional block 2180 (FIG. 21). The current series of binarized TEM tilt images corresponds to a series of TEM tilt images observed for the carbon black aggregate. In some examples, the computing device may determine that the update criteria are not met, such that the computing device may determine that the current series of binarized TEM tilt images should not be updated. Such a negative determination ("no" branch) results in the flow of the example method 2100 being directed to block 2110 of the example method 2200 to implement the next reconstruction step. In other examples, the computing device may determine that the update criteria are met, such that the computing device may determine that the current series of binarized TEM tilt images should be updated. Such a positive determination at block 2205 (the "yes" branch) causes the flow of the example method 2200 to be directed to block 2210, where the computing device may acquire a binarized TEM image of the carbon black aggregate at the defined tilt angle. The binarized TEM image is part of a current series of binarized TEM tilt images and thus corresponds to the defined tilt angle.
[0099] At block 2215, the computing device may determine or otherwise generate a projection of the arrangement of spheres representing the current 3D model of the carbon black aggregates, as described herein in connection with the example method 2100.
[0100] In block 2220, the computing device can apply a displacement to the binarized TEM image. To that end, the displacement can be represented by a displacement vector d for a set of multiple displacement vectors. For example, each displacement vector in the set of multiple displacement vectors can be expressed as (2N L +1) × (2N L +1) pixels, where L is a natural number. Such a displacement vector d can be defined in a Cartesian coordinate system with the origin at the geometric center of the square lattice. Therefore, the displacement vector is expressed as a pair (n x , n y ), where n x and n y are in the interval [-N L , N L ]. The geometric center is located at pixel . Thus, applying the displacement may include displacing the binarized TEM image by a displacement vector representing the displacement. Applying the displacement results in a displaced binarized TEM image.
[0101] At block 2225, the computing device may determine a goodness of fit metric based on the displaced binarized TEM image and the projection. The goodness of fit metric may be the same goodness of fit metric described herein in connection with block 2140 of the exemplary method 2100. More specifically, the goodness of fit metric may be the number (or, in some cases, a function of the number) of mismatched pixels resulting from pixel-by-pixel differences between the binarized TEM image and the projection. Thus, determining the goodness of fit metric may include determining the number of mismatched pixels. Although not shown, as part of the exemplary method 2200, the computing device may update a data structure to add the determined goodness of fit metric.
[0102] In block 2230, the computing device may determine whether to apply another displacement to an image in the current series of binarized TEM tilt images. If the set of multiple displacement vectors has not been completely traversed, a positive determination ("Yes" branch) may be made and the flow of the exemplary method 2200 may be directed to block 2220. In other instances where the set of multiple displacement vectors has been completely traversed, a negative determination ("No" branch) may be made and the flow of the exemplary method 2200 may be directed to block 2235, where the computing device may identify or otherwise determine a displaced binarized image that corresponds to a satisfactory goodness of fit metric. Since each goodness of fit metric determined in block 2225 indicates a number of mismatched pixels, a satisfactory goodness of fit metric may indicate a minimum number of mismatched pixels. That is, the displaced binarized image identified may have a minimum pixel mismatch with respect to the projection.
[0103] At block 2240, the computing device may update the next series of binarized TEM tilt images to include the displaced binarized images identified at block 2235. Thus, the next series of binarized TEM tilt images may be formed by the displaced binarized images that each minimize pixel mismatch with the projection determined (or otherwise generated) at block 2215.
[0104] In block 2245, the computing device can determine whether the next binarized TEM image in the current series of binarized TEM tilt images undergoes a set of multiple displacement vectors. If the current series of binarized TEM tilt images has not been completely traversed, a positive determination ("yes" branch) can be made, and the flow of the exemplary method 2200 can be directed to block 2210. In other instances where the current series of binarized TEM tilt images has been completely traversed, a negative determination ("no" branch) can be made, and the flow of the exemplary method 2200 can be directed to block 2250, where the computing device can configure the next series of binarized TEM tilt images as the current series of binarized TEM tilt images. The flow of the exemplary method 2200 can then be directed to block 2110 of the exemplary method 2100 shown in FIG. 21 to implement a further reconstruction step.
[0105] 23 illustrates an example of a method 2300 for providing a 3D reconstruction and analysis of a carbon black aggregate, according to one or more embodiments of the present disclosure. The example method 2300 can be implemented to identify the type of carbon black in a carbon black aggregate. A computing device or a system of computing devices can implement the example method 2300 in whole or in part. To that end, each of the computing devices includes computing resources capable of implementing at least one of the blocks included in the example method 2300 and other methods described herein. The computing resources include, for example, a central processing unit (CPU), a graphics processing unit (GPU), a tensor processing unit (TPU), memory, disk space, input bandwidth, and / or output bandwidth, interface(s) (such as an I / O interface or API, or both), controller device(s), power source, combinations of the foregoing, and / or similar resources. The computing device may embody the computing device 250 (FIG. 19) and thus may host at least one of the capture module 252, the segmentation module 256, the reconstruction module 260, the analysis module 270, the CB identification module 1910, and the reporting module 274. The computing device may thus implement the example method 2300 by executing one or more instances of such modules.
[0106] Exemplary method 2300 includes blocks 2010, 2020, and 2030 as illustrated in FIG. 20 and described above in connection with exemplary method 2000. A computing device implementing exemplary method 2300 may implement each of blocks 2010, 2020, and 2030 as described above. As a result of the implementation of blocks 2010, 2020, and 2030, a 3D aggregate model of the carbon black aggregate is generated. In block 2310, the computing device may use the 3D aggregate model to determine values for each of a plurality of morphological properties of the carbon black aggregate. As described above, the plurality of morphological properties may include an aggregate inequality metric (e.g., Z rel ), aggregate volume, relative 3D void volume (3DVV), or 3D specific surface area (3DSSA). As part of, or in addition to, determining such values, the computing device may generate a feature vector having a number M of components equal to the number of the plurality of morphological components whose values are determined.
[0107] In block 2320, the computing device can provide the respective values to a predictive model (e.g., predictive model 1920 (FIG. 19)). To that end, a component of the computing device (e.g., analysis module 270 (FIG. 19)) can pass the feature vector to another component (e.g., CB identification module 1910) that has the predictive model 1920.
[0108] At block 2330, the computing device may determine the type of carbon black in the carbon black aggregate by applying a predictive model to the feature vector including the respective values. In some cases, applying the predictive model may include applying a defined procedure corresponding to the prediction having the feature vector as an argument to the procedure.
[0109] In block 2340, the computing device may cause a notification of the type of carbon black to be presented. The notification may be presented on a display device operatively coupled to or integrated with the computing device, such as display device 280 (FIG. 19).
[0110] Although not shown in FIG. 23 , in some embodiments, rather than analyzing carbon black aggregates individually, the exemplary method 2300 may be implemented for a group of carbon black aggregates. To that end, blocks 2010-2030, 2310-2330 may be implemented for each carbon black aggregate in the group of carbon black aggregates. The computing device may then evaluate various predictions for each carbon black aggregate in the group of carbon black aggregates, and based on the evaluation, the computing device may determine one or more representative types of carbon black for the group of carbon black aggregates. In some cases, the computing device may rank the one or more respective types and then cause a notification or indicia to be presented indicating the ranking of the one or more respective types. The notification and / or indicia may be presented on a display device.
[0111] FIG. 24 illustrates an example of a training method 2400 for generating a predictive model, according to one or more embodiments of the present disclosure. The exemplary training method 2400 is an example of a supervised learning method, and variations of this example of the training method are described below, although other training methods can be similarly implemented to train unsupervised and / or semi-supervised machine learning (predictive) models. A computing device hosting a training module 1810 can implement the exemplary training method 2400 via the training module 1810. Such a computing device can include computing resources that can implement at least one of the blocks included in the exemplary method 2000 and other methods described herein. The computing resources include, for example, a central processing unit (CPU), a graphics processing unit (GPU), a tensor processing unit (TPU), memory, disk space, input bandwidth, and / or output bandwidth, interface(s) (such as an I / O interface or API, or both), controller device(s), power supply, combinations of the foregoing, and / or similar resources. In some cases, such a computing device can be a computing device 250 described herein.
[0112] At block 2410, the example training method 2400 can determine information determining morphological characteristics of carbon black aggregates. Such information can include one or more data sets. Each data set can include labeled baseline data.
[0113] The exemplary training method 2400 may generate a training data set and a test data set at block 2420. The training data set and the test data set may be generated by constructing labeled feature vectors, each including a value for each of a plurality of morphological properties of the carbon black aggregates. In some cases, as described herein, M morphological properties may define a feature vector. The values for each of the M morphological properties define an M-tuple that embodies an instance of the feature vector. The training data set and the test data set may be generated by randomly assigning data formatted as feature vectors to either the training data set or the test data set. In some examples, the assignment of data as training samples or test samples may not be completely random. In some examples, the training data set and the test data set may be generated using only labeled baseline data for certain features extracted from data related to the morphological properties of the carbon black aggregates. In some examples, the training data set may be generated using a majority of the labeled baseline data extracted from data related to the morphological properties of the carbon black aggregates. For example, 75% of the labeled baseline data may form the training data set and 25% of the labeled data may form the test data set. Any method or technique may be used to create the training and testing data sets.
[0114] The example training method 2400 may, for example, determine (e.g., extract, select, etc.) one or more features that can be used by a classifier to designate a carbon black aggregate as being for a particular type of carbon black, at block 2430. The example training method 2400 may determine a set of training baseline features from the training dataset.
[0115] The exemplary training method 2400 may train one or more machine learning models using one or more features at block 2440. In some examples, the machine learning models may be trained using supervised learning. In other embodiments, other machine learning techniques including unsupervised learning and semi-supervised may be employed. The machine learning models trained at block 2440 may be selected based on different criteria and / or data available in the training dataset. For example, machine learning classifiers may suffer from different degrees of bias. Thus, two or more machine learning models may be trained at block 2440 and optimized, improved, and cross-validated at 2450.
[0116] The exemplary training method 2400 may select one or more machine learning models to build a prediction engine (e.g., a machine learning classifier, a predictive model, etc.) at 2460. The prediction engine may be evaluated using a test data set. The prediction engine may analyze the test data set and generate a classification value and / or a prediction value at block 2470. The classification and / or prediction value may be evaluated at block 2480 to determine whether such values have achieved a desired level of accuracy. The performance of the prediction engine may be evaluated in some manner based on true positive, false positive, true negative, and / or false negative classification of some of the data points indicated by the prediction engine. For example, a false positive of the prediction engine may refer to the number of times the prediction engine erroneously determines exposure and / or risk of exposure to an airborne pathogen, etc. Conversely, a false negative of the prediction engine may refer to the number of times the machine learning model inaccurately determines exposure and / or risk of exposure to an airborne pathogen, etc. when, in fact, the determined exposure and / or risk of exposure to the airborne pathogen, etc., matches the actual exposure and / or risk of exposure. True negatives and true positives can refer to the number of times the predictive engine correctly determines, for example, exposure and / or risk of exposure to an airborne pathogen. Related to these measurements are the concepts of recall and precision. Broadly speaking, recall refers to the ratio of true positives to the sum of true positives and false negatives, and quantifies the sensitivity of the predictive engine. Similarly, precision refers to the ratio of true positives to the sum of true positives and false positives.
[0117] If such a desired level of accuracy is reached ("yes" branch), the training phase ends and the prediction engine can be output at 2490. If the desired level of accuracy is not reached ("no" branch), a subsequent iteration of the example training method 2400 can be performed beginning at block 2410, e.g., with modifications such as considering a larger set of data regarding morphological properties of carbon black aggregates.
[0118] The 3D analysis of carbon black according to the embodiments described herein can be implemented on a computing system 2500 shown in FIG. 25 and described below. The computer-implemented methods, devices, and systems disclosed herein can utilize one or more computing devices to perform one or more functions at one or more locations. FIG. 25 is a block diagram illustrating an exemplary computing system 2500 for performing the disclosed methods and / or implementing the disclosed systems. The computing system 2500 is only one example of an operating environment and is not intended to suggest any limitation as to the scope of use or functionality of the operating environment architecture. Neither should the operating environment be interpreted as having any dependency or requirement relating to any one or combination of components illustrated in the exemplary operating environment. The computing environment 2500 shown in FIG. 25 can embody or constitute the exemplary operating environment 200 (FIG. 2). The computing system 2500 can implement various functions described herein in connection with the TEM-based 3D reconstruction and analysis of carbon black aggregates. For example, one or more of the computing devices forming the computing system 2500 may include the capture module 252, the segmentation module 256, the reconstruction module 260, the analysis module 270, and the reporting module 274. Additionally, or in some embodiments, as described herein, one or more of the computing devices forming the computing system 2500 may also optionally include a CB identification module 1910. Additionally, or in other embodiments, one or more of the computing devices forming the computing system 2500 may also optionally include a training module 1810.
[0119] The computer-implemented methods, devices, and systems according to the present disclosure may be operational with numerous other general purpose or special purpose computing system environments or configurations. Examples of well-known computing systems, environments, and / or configurations that may be suitable for use with the systems and methods include, but are not limited to, personal computers, server computers, laptop devices, and multiprocessor systems. Additional examples include set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, distributed computing environments that include any of the above systems or devices, and the like.
[0120] The processing of the disclosed computer-implemented methods, devices, and systems may be performed by software components. The disclosed systems, devices, and computer-implemented methods may be described in the general context of computer-executable instructions, such as program modules, executed by one or more computers or other devices. Generally, program modules include computer code, routines, programs, objects, components, data structures, etc. that perform particular tasks or implement particular abstract data types. The disclosed methods may also be practiced in grid-based and distributed computing environments where tasks are performed by remote processing devices linked through a communications network. In a distributed computing environment, program modules may be located in both local and remote computer storage media, including memory storage devices.
[0121] Additionally, the systems, devices, and computer-implemented methods disclosed herein may be implemented via a general purpose computing device in the form of a computing device 2501. Components of the computing device 2501 may include one or more processors 2503, a main memory 2512, and a system bus 2513 that couples various system components, including the one or more processors 2503, to the main memory 2512. The system may utilize parallel computing. The computing device 2501 may embody a computing device 250 described herein.
[0122] The system bus 2513 represents one or more of several possible types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, or a local bus using any of a variety of bus architectures. The system bus 2513, and all buses specified herein, may also be implemented via a wired or wireless network connection, and each of the subsystems, including one or more processors 2503, mass storage device 2504, operating system 2505, software 2506, data 2507, network adapter 2508, system memory 2512, input / output interface 2510, display adapter 2509, display device 2511, and human machine interface 2502, may be contained within one or more remote, physically separate computing devices 2514a,b,c connected via this form of bus, in effect implementing a completely distributed system.
[0123] Computing device 2501 typically includes a variety of computer readable media. Exemplary readable media may be any available media accessible by computing device 2501, including, for example, both volatile and non-volatile, removable and non-removable media. Main memory 2512 includes computer readable media in the form of volatile memory, such as random access memory (RAM), and / or non-volatile memory, such as read only memory (ROM). Main memory 2512 typically includes data, such as data 2507, and / or program modules, such as operating system 2505 and software 2506, that are immediately accessible to and / or currently being operated by one or more processors 2503. For example, software 2506 may include capture module 252, segmentation module 256, reconstruction module 260, analysis module 270, and reporting module 274. Additionally or in other embodiments, software 2506 may include CB identification module 1910 or training module 1810, or both. The operating system 2505 may be embodied, for example, in one of the Windows operating systems, Unix, or Linux.
[0124] In another aspect, the computing device 2501 may also include other removable / non-removable, volatile / non-volatile computer storage media. For example, FIG. 25 illustrates a mass storage device 2504 that may provide non-volatile storage of computer code, computer readable instructions, data structures, program modules, and other data for the computing device 2501. For example, the mass storage device 2504 may be, but is not limited to, a hard disk, a removable disk, a removable optical disk, a magnetic cassette or other magnetic storage device, a flash memory card, a CD-ROM, a digital versatile disk (DVD) or other optical storage, a random access memory (RAM), a read only memory (ROM), an electrically erasable programmable read only memory (EEPROM), and the like.
[0125] Any number of program modules may be stored on the mass storage device 2504, including, by way of example, an operating system 2505 and software 2506. Each of the operating system 2505 and software 2506 (or some combination thereof) may include elements of programming and software 2506. Data 2507 may also be stored on the mass storage device 2504. The data 2507 may be stored in any of one or more databases known in the art. Examples of such databases include DB2®, Microsoft® Access, Microsoft® SQL Server, Oracle®, mySQL, PostgreSQL, SQLite, etc. The databases may be centralized or distributed across multiple systems.
[0126] In another aspect, a user can input commands and information into the computing device 2501 through input devices (not shown). Examples of such input devices include, but are not limited to, keyboards, pointing devices (e.g., a "mouse"), microphones, joysticks, scanners, tactile input devices such as gloves, and other body coverings. These and other input devices can be connected to the one or more processors 2503 through a human-machine interface 2502 coupled to the system bus 2513, but can also be connected by other interface and bus structures such as a parallel port, a game port, an IEEE 1394 port (also known as a Firewire port), a serial port, or a universal serial bus (USB).
[0127] In yet another aspect, a display device 2511 may also be connected to the system bus 2513 via an interface such as a display adapter 2509. It is contemplated that the computing device 2501 may have more than one display adapter 2509 and that the computing device 2501 may have more than one display device 2511. For example, the display device 2511 may be a monitor, an LCD (liquid crystal display), a projector. In addition to the display device 2511, other output peripheral devices may include components such as speakers (not shown) and a printer (not shown) that may be connected to the computing device 2501 via the input / output interface 2510. Any operation and / or result of the method may be output to an output device in any form. Such output may be any form of visual representation, including but not limited to text, graphical, animation, audio, tactile, etc. The display device 2511 and the computing device 2501 may be part of a single device or separate devices.
[0128] The computing device 2501 may operate in a networked environment using logical connections to one or more remote computing devices 2514a,b,c. For example, the remote computing devices may be a personal computer, a portable computer, a smart phone, a server device, a router device, a network computer, a peer device or other common network node, etc. The logical connections between the computing device 2501 and the remote computing devices 2514a,b,c may be made through a network 2515, such as a LAN and / or a general WAN. Such network connections may be through a network adapter 2508. The network adapter 2508 may be implemented in both wired and wireless environments. The network 2515 may embody or constitute, for example, the network 240(s) (FIG. 2).
[0129] For purposes of illustration, application programs and other executable program components, such as the operating system 2505, are illustrated herein as separate blocks, with it being recognized that such programs and components reside at various times in different storage components of the computing device 2501 and are executed by one or more processors 2503 of the computer. An implementation of the software 2506 may be stored on or transmitted across any form of computer-readable media. Any of the disclosed methods may be implemented by computer-readable instructions embodied on a computer-readable medium. A computer-readable medium may be any available medium that can be accessed by a computer. By way of example and not limitation, computer-readable media may include "computer storage media" and "communications media." "Computer storage media" includes volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information, such as computer-readable instructions, data structures, program modules, or other data. Exemplary computer storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disks (DVD) or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to store the desired information and that can be accessed by a computer.
[0130] 26 is a schematic block diagram of an example of a computing system 2600 capable of implementing 3D analysis of carbon black in accordance with one or more embodiments of the present disclosure. The computing system 2600 can embody or include the modeling system 245 included in the operating environment 200 (FIG. 2) or the operating environment 1900 (FIG. 19). Thus, the exemplary computing system 2600 can be implemented in the operating environment 200 to provide the functionality described herein in connection with the TEM-based 3D reconstruction and analysis of carbon black aggregates described herein.
[0131] The exemplary computing system 2600 includes two types of server devices: computation server devices 2610 and storage server devices 2620. A subset of the computation server devices 2610 may host various modules that, individually or collectively, enable implementing the TEM-based 3D reconstruction and analysis of carbon black aggregates according to the aspects described herein. Such a subset of the computation server devices 2610 may thus operate according to the functionality described herein in connection with the TEM-based 3D reconstruction and analysis of carbon black aggregates. For illustrative purposes, the computation server devices 2612 in such a subset are depicted diagrammatically as hosting those modules, e.g., the capture module 252, the segmentation module 256, the reconstruction module 260, the analysis module 270, and the reporting module 274. The computation server devices 2612 may also optionally include a CB identification module 1910 (FIG. 19) or a training module 1810 (FIG. 18), or both. Similar to the computing device 2401 (FIG. 24), the architecture of the computing server device 2612 comprises one or more processors, one or more memory devices, and a bus architecture that operatively couples the processor(s) and memory device(s). In some cases, modules hosted by the computing server device 2612 may be stored in at least one of such memory device(s). At least a subset of the computing server devices 2610 may be operatively coupled to one or more of the storage server devices 2620. The coupling may be direct or may be mediated by at least one of the gateway devices 2630. The storage server device 2620 includes data and / or metadata that may be used to implement the functionality described herein in connection with the TEM-based 3D reconstruction and analysis of carbon black aggregates. Some or all of the storage server device 2620 may embody or comprise memory 278.Additionally, or in some embodiments, some or all of the storage server devices may embody or constitute the labeled data repository 1804 and the model repository 1830.
[0132] Each of the gateway devices 2630 may include one or more processors operatively coupled to one or more memory devices that may hold application programming interfaces (APIs) and / or other types of program code for access to the compute server devices 2610 and the storage server devices 2620. Such access may be programmatic, for example, via appropriate function calls. A subset of the compute server devices 2610 that hosts one or a combination of the acquisition module 252, segmentation module 256, reconstruction module 260, analysis module 270, and reporting module 274 may also use the API(s) provided by the gateway device 2630 to provide results of performing the functions described herein in connection with the TEM-based 3D reconstruction and analysis of carbon black aggregates according to aspects described herein.
[0133] It is to be understood that the methods and systems described herein are not limited to the particular operations, processes, components, or structures described, or to the ordering or specific combination of such operations or components as described, and that the terminology used herein is for the purpose of describing example embodiments only, and is not intended to be limiting or restrictive.
[0134] As used herein, the singular forms "a," "an," and "the" include both singular and plural referents unless the context clearly dictates otherwise. Values expressed as approximations, by use of antecedents such as "about" or "approximately," are intended to include a reasonable variation from the referenced value. When such approximations are included along with ranges, not only are the endpoints intended to be approximations, but the magnitude of the range is also intended to be approximations. Lists are to be considered exemplary and are not limited or restricted to the elements comprising the list or to the order in which the elements are listed, unless the context clearly dictates otherwise.
[0135] Throughout the specification and claims of this disclosure, the following terms have the meanings set forth. "Comprise" and variations of the word, such as "comprising" and "comprises," mean including, but not limited to, and do not exclude, for example, other appendants, components, integers, or operations. Variations of the word, such as "include" and "including," are not intended to mean something restricted or limited to what is shown as being included, or to exclude something not shown. "May" means something permissive, but not restrictive or limiting. "Optional" or "optionally" means something that may or may not be included without changing the outcome or what is being described. Variations of the word, such as "prefer" and "preferred" or "preferably" mean something that is exemplary and more ideal, but not required. "Such as" means something that merely serves as an example.
[0136] The operations and components described herein as being used to perform the disclosed methods and to construct the disclosed systems are exemplary unless the context clearly dictates otherwise. When combinations, subsets, interactions, groups, etc. of these operations and components are disclosed, it should be understood that although specific references to various individual and collective combinations and permutations of each of these may not be expressly disclosed, each is specifically contemplated and described herein for all methods and systems. This applies to all aspects of this application, including, but not limited to, the operations in the disclosed methods and / or components disclosed in the systems. Thus, where there are various additional operations that may be performed or components that may be added, it should be understood that each of these additional operations may be performed and components may be added with any particular embodiment or combination of embodiments of the disclosed systems and methods.
[0137] The embodiments of the present disclosure may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the method and system may take the form of a computer program product on a computer-readable storage medium having computer-readable program instructions (e.g., computer software) embodied in the storage medium. Any suitable computer-readable storage medium, whether internal, networked, or cloud-based, may be utilized, including hard disks, CD-ROMs, optical storage devices, magnetic storage devices, memregisters, non-volatile random access memory (NVRAM), flash memory, or combinations thereof.
[0138] The embodiments of the present disclosure are described with reference to diagrams, flowcharts, and other illustrations of computer-implemented methods, systems, devices, and computer program products. Each block of the block diagrams and flowchart diagrams, and combinations of blocks in the block diagrams and flowchart diagrams, respectively, may be implemented by processor-accessible instructions. Such instructions may include, for example, computer program instructions (e.g., processor-readable instructions and / or processor-executable instructions). The processor-accessible instructions may be constructed (e.g., linked and compiled) and stored in a processor-executable form in one or more memory devices or one or many other processor-accessible non-transitory storage media. These computer program instructions (constructed or not) may be loaded into a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine. The loaded computer program instructions may be accessed and executed by one or more processors or other types of processing circuitry. In response to execution, the loaded computer program instructions provide the functionality described in association with the flowchart blocks (individually or in a specific combination) or the blocks in the block diagrams (individually or in a specific combination). Thus, such instructions, executed on a computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart blocks (individually or in a specific combination) or in the blocks in the block diagrams (individually or in a specific combination).
[0139] These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory generate an article of manufacture that includes processor-accessible instructions (e.g., processor-readable instructions and / or processor-executable instructions) for implementing the functions specified in the flowchart blocks (individually or in a specific combination) or block diagram blocks (individually or in a specific combination). The computer program instructions (constructed or otherwise) may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operations to be performed on the computer or other programmable apparatus, generating a computer-implemented process. The series of operations may be performed in response to execution by one or more processors or other types of processing circuits. Thus, such instructions executing on a computer or other programmable apparatus provide operations for implementing the functions specified in the flowchart blocks (individually or in a specific combination) or block diagram blocks (individually or in a specific combination).
[0140] Thus, the blocks of the block diagrams and flowchart diagrams support combinations of means for performing the functions specified in connection with such diagrams and / or flowchart diagrams, combinations of operations for performing the specified functions, and program instruction means for performing the specified functions. Each block of the block diagrams and flowchart diagrams, and combinations of blocks in the block diagrams and flowchart diagrams, may be implemented by a dedicated hardware-based computer system that performs the specified functions or operations, or a combination of dedicated hardware and computer instructions.
[0141] As used herein and in the accompanying drawings, terms such as "module," "component," "system," "platform," and the like, may refer to and / or include computer-related entities or entities related to an operable machine having one or more particular functions. Such entities may be either hardware, a combination of hardware and software, software (e.g., program code or executable program code), or software in execution. In one example, a component may be a process running on a processor, a processor, an object, an executable file (e.g., binary software), a thread of execution, a computer program, and / or a computing device. By way of example only, a software application running on a server device may be a component, and the server device may also be a component. One or more modules may reside within a process and / or thread of execution. One or more components may also reside within a process and / or thread of execution. Each of the modules and components may be localized on one computing device and / or distributed among two or more computing devices. In another example, each component (or module) may execute from different computer-readable storage media having different data structures stored thereon. Components (or modules) may communicate via local and / or remote processes, such as following a signal having one or more data packets (e.g., data from one component interacting with another component in a local system, a distributed system, and / or other systems via a network such as the Internet via a signal). As another example, in some cases, a component may emulate an electronic component via a virtual machine connection, for example, in a cloud computing system. The terms "module" and "component" (and their plurals) may be used interchangeably in some cases, where clear from the context.
[0142] As used herein and in the accompanying drawings, the term "processor" may refer to substantially any computing processing unit or device, including single-core processors, single processors with software multithreading execution capabilities, multi-core processors, multi-core processors with software multithreading execution capabilities, multi-core processors with hardware multithreading techniques, parallel platforms, and parallel platforms with distributed shared memory. Additionally, a processor may refer to electronic circuits assembled and designed to execute code instructions and / or operate on data and signaling. Such electronic circuits may be incorporated, for example, into a chipset. Thus, in some cases, a processor may be embodied in or include application specific integrated circuits (ASICs), digital signal processors (DSPs), field programmable gate arrays (FPGAs), complex programmable logic devices (CPLDs), discrete gate or transistor logic, discrete hardware components, or any combination thereof designed and assembled to perform the functions described herein. Additionally, in some cases, a processor may utilize nanoscale architectures, such as molecular and quantum dot-based transistors, switches, and gates, to optimize space usage or enhance the performance of a computing device. A processor may also be implemented as a combination of computing processing units.
[0143] Additionally, in this specification and the accompanying drawings, terms such as "storage," "data storage," "repository," and substantially any other information storage component associated with the operation and functionality of systems, subsystems, modules, and components are utilized to refer to "memory components," entities incorporated in the "memory," or components that include memory. As described herein, the memory and / or memory components of the present disclosure may be either volatile or non-volatile memory, or may include both volatile and non-volatile memory. By way of example only, non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable ROM (EEPROM), flash memory, or non-volatile random access memory (RAM) (e.g., ferroelectric RAM (FeRAM)). Volatile memory may include, for example, RAM, which may function as an external cache memory. By way of example and not limitation, RAM is available in many forms, such as Synchronous RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDR SDRAM), Enhanced SDRAM (ESDRAM), Synchlink DRAM (SLDRAM), Direct Rambus RAM (DRRAM), Direct Rambus Dynamic RAM (DRDRAM), and Rambus Dynamic RAM (RDRAM), etc. Embodiments of the present disclosure are not limited to these types of memory and other types of memory devices may be contemplated.
[0144] The methods and systems may employ artificial intelligence techniques such as machine learning and iterative learning. Examples of such techniques include, but are not limited to, expert systems, case-based reasoning, Bayesian networks, behavior-based AI, neural networks, fuzzy systems, evolutionary computing (e.g., genetic algorithms), swarm intelligence (e.g., ant algorithms), and hybrid intelligent systems (e.g., expert inference rules generated through neural networks or generative rules from statistical learning).
[0145] Although the computer-implemented methods, apparatus, devices, and systems have been described in connection with preferred embodiments and specific examples, the embodiments herein are intended in all respects to be illustrative and not restrictive, and therefore the scope is not intended to be limited to the particular embodiments described.
[0146] Unless expressly stated otherwise, it is never intended that any method described herein be construed as requiring that its operations be performed in a particular order. Thus, if a method claim does not actually recite an order in which its operations should be followed, or if it is not otherwise specifically stated in the claim or description that the operations are to be limited to a particular order, no order is ever intended to be inferred in any respect. This applies to any possible non-express basis for interpretation, including questions of logic regarding the arrangement of operations or operation flow, plain meaning derived from grammatical constructions or punctuation, or the number or type of embodiments described herein.
[0147] It will be apparent to those skilled in the art that various modifications and variations can be made without departing from the scope or spirit of the invention. Other embodiments will be apparent to those skilled in the art from consideration of the specification and practice disclosed herein. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit being indicated by the following claims.< / d> < / d>
Claims
1. 1. A computer-implemented method comprising: Identifying carbon black aggregates using two-dimensional (2D) transmission electron microscope (TEM) images; determining a number of spheres based on at least an approximation of a volume of the carbon black agglomerates; generating a three-dimensional (3D) agglomerate model of the carbon black agglomerate by determining a solution to an optimization problem relating to an objective function based on the sizes of the spheres and the positions of the spheres within a defined volume, the solution being representative of the 3D agglomerate model; 4. A computer-implemented method comprising:
2. The computer-implemented method of claim 1 , further comprising: causing a notification to be presented that the 3D aggregate model is available for analysis.
3. determining values of each of one or more morphological properties of the carbon black aggregates using the 3D aggregate model; and The computer-implemented method of claim 1 , further comprising: providing the respective values of the one or more morphological properties.
4. 4. The computer-implemented method of claim 3, wherein the one or more morphological properties include at least one of an aggregate anisotropy metric, an aggregate volume, a relative 3D void volume (3DVV), or a 3D specific surface area (3DSSA).
5. The providing comprises: retaining data indicative of the one or more morphological characteristics; configuring an application programming interface to access a value of a first morphological property of the one or more morphological properties; The computer-implemented method of claim 3 , comprising:
6. Determining the solution to the optimization problem comprises: updating a current arrangement of the spheres, the current arrangement of the spheres including a distribution of sizes of the spheres and a distribution of positions of the spheres within the defined volume; determining a plurality of 2D projections of said configuration of spheres onto respective defined planes; The computer-implemented method of claim 1 , comprising:
7. The updating step comprises: modifying a position of a first one of the spheres and maintaining a second position of a second one of the spheres; or modifying a size of the first sphere and maintaining a second size of the second sphere; The computer-implemented method of claim 6 , comprising at least one of:
8. 7. The computer-implemented method of claim 6, wherein determining the solution to the optimization problem further comprises determining a plurality of goodness of fit metrics for each of the plurality of 2D projections, each goodness of fit metric of the plurality of goodness of fit metrics being based on a respective binarized 2D image of the carbon black aggregate.
9. 9. The computer-implemented method of claim 8, wherein the determining the solution to the optimization problem further comprises updating the objective function based on an average of the plurality of fitness metrics.
10. determining that the objective function satisfies a convergence criterion; constructing the current arrangement of spheres as the 3D aggregate model; The computer-implemented method of claim 9 , further comprising:
11. 1. A computing device comprising: one or more processors; and one or more memory devices storing computer-executable instructions that, in response to execution by the one or more processors, cause the computing device to: Identifying carbon black aggregates using two-dimensional (2D) transmission electron microscope (TEM) images; determining a number of spheres based at least on an approximation of a volume of the carbon black aggregates; and generating a three-dimensional (3D) agglomerate model of the carbon black agglomerate by determining a solution to an optimization problem relating to an objective function based on the sizes of the spheres and the positions of the spheres within a defined volume, the solution being representative of the 3D agglomerate model; A computing device that causes
12. 12. The computing device of claim 11, wherein the one or more memory devices store further computer-executable instructions that, in response to execution by the one or more processors, further cause the computing device to provide a notification that the 3D aggregate model is available for analysis.
13. The one or more memory devices, in response to execution by the one or more processors, determining values of each of one or more morphological properties of the carbon black aggregates using the 3D aggregate model; and providing said respective values of said one or more morphological properties; 12. The computing device of claim 11 further storing further computer-executable instructions to further cause the computing device to:
14. Determining the solution to the optimization problem comprises: updating a current arrangement of the spheres, the current arrangement of the spheres including a distribution of sizes of the spheres and a distribution of positions of the spheres within the defined volume; determining a plurality of 2D projections of said configuration of spheres onto respective defined planes; The computing device of claim 11 , comprising:
15. 15. The computing device of claim 14, wherein the determining the solution to the optimization problem further comprises determining a plurality of goodness of fit metrics for each of the plurality of 2D projections, each goodness of fit metric of the plurality of goodness of fit metrics being based on a respective binarized 2D image of the carbon black aggregate.
16. The computing device of claim 15 , wherein the determining the solution to the optimization problem further comprises updating the objective function based on an average of the plurality of fitness metrics.
17. The one or more memory devices, in response to execution by the one or more processors, determining that the objective function satisfies a convergence criterion; constructing the current arrangement of spheres as the 3D aggregate model; 20. The computing device of claim 16 further storing further computer-executable instructions to further cause the computing device to:
18. At least one non-transitory computer-readable storage medium having processor-executable instructions encoded thereon, the processor-executable instructions, in response to execution, causing a computing device to: Identifying carbon black aggregates using two-dimensional (2D) transmission electron microscope (TEM) images; determining a number of spheres based at least on an approximation of a volume of the carbon black aggregates; and generating a three-dimensional (3D) agglomerate model of the carbon black agglomerate by determining a solution to an optimization problem relating to an objective function based on the sizes of the spheres and the positions of the spheres within a defined volume, the solution being representative of the 3D agglomerate model; At least one non-transitory computer-readable storage medium that causes
19. 20. The at least one non-transitory computer-readable storage medium of claim 18, wherein the processor-executable instructions, upon further execution, further cause the computing device to provide a notification that the 3D aggregate model is available for analysis.
20. The processor-executable instructions, in response to further execution, cause the computing device to: determining values of each of one or more morphological properties of the carbon black aggregates using the 3D aggregate model; and providing said respective values of said one or more morphological properties; 20. The at least one non-transitory computer-readable storage medium of claim 18, further comprising:
21. Determining the solution to the optimization problem comprises: updating a current arrangement of the spheres, the current arrangement of the spheres including a distribution of sizes of the spheres and a distribution of positions of the spheres within the defined volume; determining a plurality of 2D projections of said configuration of spheres onto respective defined planes; 20. At least one non-transitory computer-readable storage medium according to claim 18, comprising:
22. 22. The at least one non-transitory computer-readable storage medium of claim 21 , wherein the determining the solution to the optimization problem further comprises determining a plurality of goodness of fit metrics for each of the plurality of 2D projections, each goodness of fit metric of the plurality of goodness of fit metrics being based on a respective binarized 2D image of the carbon black aggregate.
23. 23. The at least one non-transitory computer-readable storage medium of claim 22, wherein the determining the solution to the optimization problem further comprises updating the objective function based on an average of the plurality of fitness metrics.
24. The one or more memory devices, in response to execution by the one or more processors, determining that the objective function satisfies a convergence criterion; constructing the current arrangement of spheres as the 3D aggregate model; 24. At least one non-transitory computer-readable storage medium according to claim 23 storing further computer-executable instructions further causing the computer to:
Citation Information
Patent Citations
Method for preparing solid amorphous alloy thin strips with controllable microstructures
CN105624587A
Form display device of rubber material and form display method of the rubber material
JP2006200938A
TEM-based metrology method and system
US20190393016A1
Filler structure retention inpolymeric compositions
WO2021168422A1
