Method and equipment for analyzing particles in cigarette filter tip

Abstract

The invention relates to a method and equipment for analyzing particles in a cigarette filter tip, and the method comprises the steps: collecting data information of a cigarette from different angles in the circumferential direction through CT equipment, and constructing a three-dimensional reconstruction model of the cigarette based on the data information; splitting a particle region from the three-dimensional reconstruction model according to gray difference to form a point cloud data set of the particle region; performing sphere fitting on the point cloud data set of the particle area to obtain the sphere center of each fitted sphere; calculating the distances between all the inner points of each fitted sphere and the sphere center, and accumulating the numbers of the inner points with the distances different from the corresponding radiuses to obtain the number of abnormal inner points; when the number of the abnormal inner points of the fitted sphere reaches the preset inner point number threshold value for judging whether the particle surface is smooth or not, it is determined that the particle surface corresponding to the fitted sphere is not smooth, the to-be-detected cigarette does not need to be disassembled, a three-dimensional model does not need to be constructed, the analysis precision and efficiency are improved, and more economic benefits are brought.

Description

Technical Field

[0001] This invention relates to a method and apparatus for analyzing particles in cigarette filters, belonging to the field of cigarette product testing technology. Background Technology

[0002] Currently, the relatively monotonous flavor characteristics of traditional cigarettes can no longer fully satisfy consumers' sensory needs for cigarette products. The cigarette industry has designed new products that add particles to cigarette filters to enhance the sensory richness of cigarette smoking. This product loads flavoring components onto particle additives and adds particles to the cigarette filter to give cigarettes different flavors, thereby meeting consumers' diverse smoking needs. The product's characteristic is that the flavoring components on the particle additives are slowly released, providing a long-lasting and rich aroma during the smoking process. At the same time, the particle additives in the cigarette filter can absorb some harmful components in mainstream smoke. The design of the particles in the filter directly affects the smoking experience of the cigarette. To analyze the design of the particles in the filter more deeply and directly (e.g., particle surface smoothness, particle sphericity, particle diameter, etc.), current technologies still rely on manual visual observation or using computer vision technology to collect two-dimensional images of the particles in the cigarette filter for analysis. Obviously, the aforementioned methods all have drawbacks such as low detection accuracy and low precision.

[0003] To address the aforementioned issues, patent publication CN113441410B discloses a method and system for detecting the quality of cigarette flavor capsules. This method involves continuously acquiring image data of the flavor capsule during its spiral rolling process. The image data includes the diameter of the flavor capsule at different positions during the spiral rolling process. The method compares the diameters of multiple sets of flavor capsules at mutually axially symmetrical positions. If the differences are all less than or equal to a third preset value, the product is deemed to have passed quality inspection. Alternatively, the method compares the diameters of multiple sets of flavor capsules at mutually centrally symmetrical positions. If the differences are all less than or equal to a fourth preset value, the product is deemed to have passed quality inspection. As can be seen, in this patent document, the diameters of images obtained from any two mutually symmetrical positions during the rolling process of the burst beads are compared to reflect the roundness and shape uniformity of the burst beads. However, if all the burst beads to be tested are substandard, or if any two burst beads at mutually symmetrical positions selected are substandard, then the quality analysis of the burst beads cannot be performed, or the analysis results will be incorrect. At the same time, this patent document only compares the diameters of burst beads at symmetrical positions, without obtaining specific parameters for each burst bead (e.g., whether the surface is smooth, particle sphericity, particle diameter, etc.), so the person being tested still cannot gain a deep understanding of each burst bead. Summary of the Invention

[0004] The purpose of this invention is to provide a method and apparatus for analyzing particles in cigarette filters, which solves the technical problem of not being able to directly and efficiently analyze the relevant parameters of particles added to filters.

[0005] To achieve the above objectives, on the one hand, the present invention proposes a method for analyzing particles in cigarette filters, comprising:

[0006] Data information of cigarettes was collected from different circumferential angles using CT equipment, and a three-dimensional reconstruction model of the cigarettes was constructed based on the data information;

[0007] The granular regions are split from the 3D reconstruction model according to grayscale differences to form a point cloud dataset of the granular regions.

[0008] Perform sphere fitting on the point cloud dataset of the granular region to obtain the center of each fitted sphere;

[0009] For each fitted sphere, the distance from all its interior points to the center of the sphere is calculated, and the number of interior points whose distances are different from their corresponding radii is accumulated to obtain the number of abnormal interior points. When the number of abnormal interior points of a fitted sphere reaches the preset threshold for the number of interior points used to determine whether the particle surface is smooth, it is determined that the particle surface corresponding to the fitted sphere is not smooth.

[0010] Furthermore, the open3D open-source library was used to process the point cloud data contained in each fitted sphere to obtain the corresponding surface area and volume; based on the surface area and volume of each fitted sphere, the sphericity index of the fitted sphere was calculated.

[0011] The triangular mesh model algorithm is used to calculate the surface area, and the voxel mesh algorithm is used to calculate the volume.

[0012] Furthermore, the sphericity index is calculated using the following formula:

[0013]

[0014] Where V is the volume of the particle; S is the surface area of ​​the particle.

[0015] Furthermore, the radius of each fitted sphere is calculated to obtain the average diameter of the particle.

[0016] Further, perform the following steps on the point cloud dataset of the granular region until the number of interior points of all fitted spheres equals the number of interior points of the point cloud dataset of the granular region, thus obtaining all fitted spheres of the granular region:

[0017] Select a preset number of data points from the point cloud dataset of the granular region, perform sphere fitting on the selected data points, and obtain the sphere corresponding to the selected data points when the fitting parameters of the fitted sphere are greater than the preset fitting parameter threshold.

[0018] Furthermore, granular regions are obtained from the 3D reconstructed model based on grayscale differences using the following method:

[0019] Based on the data of the cigarette, a training model is generated to segment the particle regions according to the grayscale differences, and a training model for extracting particle regions is generated; the training model is called on the 3D reconstruction model to segment and obtain the particle regions.

[0020] Furthermore, a filtered back projection algorithm is used to process the data information of the cigarette to generate a three-dimensional reconstruction model of the cigarette.

[0021] On the other hand, the present invention also proposes an analysis device for particles in cigarette filters, the device including a processor for executing the above-described analysis method for particles in cigarette filters.

[0022] The beneficial effects of this invention are as follows: Data information of cigarettes collected from different circumferential angles using CT equipment is used to construct a three-dimensional reconstruction model of the cigarettes based on the data information; particle regions are segmented from the three-dimensional reconstruction model according to grayscale differences to form a point cloud dataset of the particle regions; spherical fitting is performed on the point cloud dataset of the particle regions to obtain the center of each fitted sphere; the distance from all interior points of each fitted sphere to the center of the sphere is calculated, and the number of interior points whose distances differ from their corresponding radii is accumulated to obtain the number of abnormal interior points; when the number of abnormal interior points of a fitted sphere reaches a preset threshold for judging whether the particle surface is smooth, the particle surface corresponding to that fitted sphere is determined to be unsmooth; by fusing the three-dimensional reconstruction model and spherical fitting, the number of abnormal interior points within the sphere is judged for each particle in the filter, thus ensuring the smoothness of all particle surfaces. This not only eliminates the need to disassemble the cigarettes to be tested, but also constructs a three-dimensional model, improving the accuracy and efficiency of analysis and bringing greater economic benefits. Attached Figure Description

[0023] Figure 1 This is a flowchart illustrating a method for analyzing particles in cigarette filters proposed in this invention.

[0024] Figure 2 This is a schematic diagram of the operation of a CT device when collecting data information from a cigarette filter, in a practical application scenario, based on the particle analysis method for cigarette filters proposed in this invention.

[0025] Figure 3This is a schematic diagram illustrating the principle of the analysis method for particles in cigarette filters proposed in this invention during the scanning process of cigarettes using CT equipment in a practical application scenario.

[0026] Figure 4 This is a schematic diagram showing the particle content detection results of the particle analysis method for cigarette filters proposed in this invention in a practical application scenario.

[0027] Figure label:

[0028] 1-X-X-ray source; 2-Platform; 3-Sample base; 4-Sample holder; 5-Detection platform. Detailed Implementation

[0029] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments.

[0030] The inventive concept of this invention is to utilize a three-dimensional reconstruction model of the cigarette to be tested, and combine spherical fitting to transform each particle in the particle region into a sphere. Then, for each sphere, the surface smoothness, sphericity index, and diameter are calculated. This digitizes the abstract particle analysis problem, avoids the problem of not being able to directly and efficiently analyze the relevant parameters of the particles added to the filter, and is conducive to the continuous optimization of particles in industrial production and the continuous development of cigarettes.

[0031] Method Example 1:

[0032] like Figure 1 The diagram shown is a flowchart illustrating a method for analyzing particles in cigarette filters proposed in this invention, which includes steps S11-S14, specifically:

[0033] Before step S11, proceed as follows Figure 2 As shown, data on cigarettes is collected, specifically... Figure 2This is a schematic diagram illustrating the operation of a CT scanner in a practical application scenario of the method for analyzing particles in cigarette filters proposed in this invention, where the scanner collects data from the cigarette. The cigarette is fixed in place using a sample holder 4, and its position is adjusted to ensure it is perpendicular to the sample base 3. The industrial CT scanner is then opened, and the sample base 3 is placed into the mounting platform 2 within the scanner, ensuring that the clamping devices (sample base 3 and sample holder 4) are fixed to the mounting platform 2 to prevent the sample from falling off during platform 2 rotation. The X-ray source 1 is configured with an X-ray source tube voltage of 100 kV, an X-ray source tube current of 70 μA, a scanning thickness of 0.004 mm, a scanning interval of 0.004 mm, and a cone-beam CT scanning mode. Normal scanning mode; the worktable moves the carrier platform 2 so that the cigarette to be detected is located in the center of the X-ray scanning range. By controlling the rotation of the carrier platform 2, it is ensured that the cigarette is in the X-ray scanning position within a 360° range. The held cigarette is removed, and the CT equipment is calibrated with air. Then, the central axis calibration rod is placed on the carrier platform 2 for central axis calibration. The cigarette is then placed back on the carrier platform 2, and the CT scan is started to scan the cigarette. The data information received by the detection platform 5 is transmitted to the computer for storage, completing the acquisition of cigarette data information. This achieves comprehensive acquisition of cigarette data information without damaging the cigarette, providing a comprehensive and accurate data source for subsequent analysis of cigarette content.

[0034] After collecting data on the cigarette using a CT scanner, step S11 is executed. The CT scanner collects data on the cigarette from different circumferential angles and constructs a three-dimensional reconstruction model of the cigarette based on the data. Here, "different axial angles" means that data on the cigarette is collected at every axial angle around the cigarette to achieve full-angle data acquisition. Based on the comprehensive data, a three-dimensional reconstruction model of the cigarette is constructed, achieving a high degree of fidelity in the three-dimensional reconstruction model.

[0035] It should be noted that in practical applications, the filtered back projection algorithm is used to process the data information of the cigarette to generate a three-dimensional reconstruction model of the cigarette. Specifically, the filtered back projection algorithm (i.e., the FDK algorithm) is used to process the data information through the following steps: First, the acquired data is preprocessed. Images are acquired at a rate of 1° rotation, resulting in 360 cigarette images. The two-dimensional projection data acquired at each angle is weighted to correct the cone beam. Second, the corrected and weighted projection data is filtered in one dimension along a projection perpendicular to the flat panel detector. Third, the data from the second step is back-projected to reconstruct the image. Back-projection calculations are performed along the X-ray direction on the filtered data. Simultaneously, due to the large spacing in the Z-axis direction during sampling, interpolation calculations are required in the Z-axis direction. This invention uses bilinear interpolation to perform interpolation in both the x-axis and y-axis directions, thus constructing the three-dimensional reconstruction model of the cigarette.

[0036] Step S12: Segment the particle regions from the 3D reconstruction model according to grayscale differences to form a point cloud dataset of particle regions. Here, the particle regions are obtained from the 3D reconstruction model according to grayscale differences using the following method: Training is performed on the particle region segmentation according to grayscale differences based on the data information of the cigarettes to generate a training model for extracting particle regions; the training model is called on the 3D reconstruction model to segment the particle regions. By performing multiple training sessions on particle region segmentation according to grayscale differences, an accurate segmentation model is formed, which improves the accuracy of segmenting particle regions from the 3D reconstruction model and lays an accurate data foundation for subsequent data processing.

[0037] Specifically, the segmentation training is performed using the following method: A computer image processing system is used to segment images containing particles from the acquired two-dimensional image sequence. Based on the different grayscale values ​​of the particles and the filter, the particles and the filter are labeled separately in the three-view interface of the image processing software (to ensure classification accuracy, multiple two-dimensional images need to be labeled during this process). When the particle region is segmented on the 3D reconstruction model, a grayscale-based Region of Interest (ROI) is created in the image processing system. The filter region is then separated from the obtained 3D reconstruction model to form a new ROI. Next, a new ROI is formed within the filter region's ROI to extract the particle region; the particle region is then segmented from the filter region based on the new ROI.

[0038] Step S13: Perform sphere fitting on the point cloud dataset of the particle region to obtain the center of each fitted sphere.

[0039] Step S14: Calculate the distance from all interior points to the center of each fitted sphere, and sum the number of interior points whose distances differ from their corresponding radii to obtain the number of abnormal interior points. When the number of abnormal interior points of a fitted sphere reaches a preset threshold for determining whether the particle surface is smooth, the particle surface corresponding to that fitted sphere is determined to be rough. It should be noted that the interior points refer to the multiple point cloud data points corresponding to each particle in the point cloud dataset of the particle region. By fitting a sphere to the target particle using its corresponding multiple target point cloud data points, a target fitted sphere is obtained. In this case, the interior points of the target fitted sphere are the multiple target point cloud data points. In practical application scenarios, the threshold for the number of interior points will be set according to cigarette parameters, cigarette requirements, or particle requirements, etc., and is not set here.

[0040] Furthermore, perform the following steps on the point cloud dataset of the granular region until the number of interior points of all fitted spheres equals the number of interior points of the point cloud dataset of the granular region, thus obtaining all fitted spheres of the granular region:

[0041] A preset number of data points are selected from the point cloud dataset of the particle region. A sphere is fitted to these selected data points. When the fitting parameter of the fitted sphere is greater than a preset fitting parameter threshold, the sphere corresponding to the selected data points is obtained. Here, the preset number of points refers to the number of basic data points required to fit a sphere. In practical applications, the preset number of points will be set according to the particle requirements of different cigarettes. In the preferred embodiment of this application, a preset number of 4 data points is preferred. Simultaneously, when the fitting parameter of the fitted sphere is less than the preset fitting parameter threshold, data points will continue to be selected from the point cloud dataset of the particle region until the fitting parameter of the fitted sphere is greater than the preset fitting parameter threshold. The fitting parameter threshold is used to measure whether a data point in the sphere fitting process corresponds to a data point in another sphere. By setting the fitting parameter threshold, it is ensured that all interior points contained within the fitted sphere are integrated during the fitting process, avoiding phenomena such as misfitting or unfitted interior points.

[0042] Steps S11-S14 enable a rapid and efficient analysis of the surface smoothness of each particle in the cigarette filter.

[0043] In a preferred embodiment of the present invention, the cigarette A to be detected is placed on the sample holder in a CT scanner. Data information of cigarette A is acquired from various axial angles through X-ray scanning in the CT scanner. Based on the data of cigarette A, a three-dimensional reconstruction model Model of cigarette A is constructed using the FDK algorithm. According to a pre-trained model that segments based on grayscale differences, a granular region is extracted from the three-dimensional reconstruction model Model, forming a point cloud dataset M1 of the granular region. A sphere fitting is performed on the point cloud dataset M1 of the granular region. Specifically, an iterative algorithm is used, randomly selecting 4 data points each time to fit a sphere, and comparing the number of interior points of all fitted spheres with the number of interior points in the point cloud dataset M1 of the granular region. The iteration ends when the number of interior points of all fitted spheres is equal to the number of interior points in the point cloud dataset M1 of the granular region, resulting in n spheres corresponding to the granular region, as well as the center point and radius of each sphere. For each sphere, the following operations are performed: calculate the distance r between each inner point of the target sphere and its center, compare each distance r with the radius R of the target sphere, record and accumulate the abnormal inner points where r is not equal to R, and obtain the number of abnormal inner points of the target sphere; if the number of abnormal inner points reaches the preset threshold, it is determined that the surface of the target particle corresponding to the target sphere is not smooth, thereby obtaining the surface smoothness of all particles in the filter.

[0044] Method Example 2:

[0045] This invention proposes a method for analyzing particles in cigarette filters, and also analyzes the sphericity index of the particles. Specifically, the point cloud data contained within each fitted sphere is processed using the Open3D open-source library to obtain the corresponding surface area and volume. Based on the surface area and volume of each fitted sphere, the sphericity index of the fitted sphere is calculated. Here, since each fitted sphere is not a regular sphere, the conventional formulas for calculating sphere volume and surface area cannot obtain accurate volume and surface area values. Therefore, Open3D technology is used here to obtain the surface area and volume of the fitted sphere. The Open3D open-source library provides an efficient and easy-to-use tool for 3D data processing. Specifically, a triangulation mesh model algorithm is used to calculate the surface area; a voxel mesh algorithm is used to calculate the volume, improving the calculation accuracy of the surface area and volume of the fitted sphere.

[0046] In practical applications, the sphericity index is calculated using the following formula:

[0047]

[0048] Where V is the volume of the particle; S is the surface area of ​​the particle.

[0049] When calculating the sphericity index, the ratio of the surface area of ​​an equal-volume sphere (i.e., the surface area of ​​a regular sphere with the same volume as the sphere fitted to the particle, which can be calculated from the particle's volume) to the particle's surface area is used to represent the sphericity index. This index measures the degree of similarity between the particle and the regular sphere. Typically, the sphericity of a regular sphere is set to 1; the closer the sphericity index is to 1, the more regular the sphere is after the particle is fitted.

[0050] Following the above embodiments of the present invention, for each sphere, the following operations are performed: The surface area of ​​the target sphere is calculated using the Open3D open-source library, preferably using a triangulation network model algorithm. Specifically, the point cloud data of the target sphere is triangulated to obtain a triangulated mesh model; the triangulated mesh model is converted into a point cloud model; and the surface area S of the target sphere is calculated using Open3D's built-in methods. The volume is preferably calculated using an accelerated mesh algorithm. Specifically, a voxel mesh is created, and the point cloud data of the target sphere is converted into a voxel mesh through voxelization; the diameter of the voxel mesh is calculated using Open3D's built-in methods; based on the diameter of the voxel mesh, the volume of the voxel mesh, i.e., the volume V of the target sphere, is further calculated. Based on the surface area S and volume V of the target sphere, the sphericity index formula is called to obtain the sphericity index Q of the particles corresponding to the target sphere, and thus the sphericity index of all particles in the filter is obtained.

[0051] Method Example 3:

[0052] The present invention proposes a method for analyzing particles in cigarette filters, and also analyzes the average diameter of the particles. Specifically, it calculates the radius of each fitted sphere to obtain the average diameter of the particles.

[0053] In practical applications, the average diameter of the particles is calculated using the following formula:

[0054]

[0055] Where D is the average diameter, R i Let be the radius of the i-th sphere; n is the sum of all fitted spheres in the particle region.

[0056] Following the above embodiments of the present invention, the average diameter of all spheres is obtained by averaging the n spheres corresponding to the obtained particle region and the sphere radius corresponding to each sphere, which is the average diameter D of the particles in the filter.

[0057] Method Example 4:

[0058] like Figure 3The diagram illustrates the principle of a CT scanner scanning a cigarette in a practical application scenario for the particle analysis method in a cigarette filter proposed in this invention. The cigarette to be detected is positioned at point O. Images are acquired via X-rays at 1° rotation intervals. A virtual detector is used to weight the two-dimensional projection data acquired at each angle to correct the cone beam. A flat panel detector is then used to perform one-dimensional filtering on the corrected and weighted projection data, thereby acquiring cigarette data information from all angles and constructing a three-dimensional reconstruction model of the cigarette.

[0059] Method Example 5:

[0060] Furthermore, in practical applications of this invention, the radius of the filter area / particle area can also be calculated based on each subset of point clouds and its fitted circle within the filter area / particle area.

[0061] like Figure 4 The diagram shows the results of particle content detection in a practical application scenario using the particle analysis method for cigarette filters proposed in this invention. The method described in this invention is applied to the cigarette to be tested, and the measured average radius of the filter area is 5.5 mm; the particle accumulation height is 1.235 mm; and the height of the particle space is 4.516 mm.

[0062] Equipment Example:

[0063] The present invention also proposes an analysis device for particles in cigarette filters. The device includes a processor for executing the analysis method for particles in cigarette filters described above. Here, the embodiment of the analysis device is described in the embodiment of the analysis method for particle mass in cigarette filters described above, and will not be repeated here.

[0064] In summary, this invention utilizes industrial CT technology to scan cigarette filters containing particles, obtaining a three-dimensional reconstructed model of the sample. The cigarette filter portion and the particle portion are separated based on grayscale values, thus obtaining a point cloud dataset of the particle portion. By processing this point cloud dataset, the particle diameter, particle sphericity index, and surface smoothness of the particles are obtained. This solves the problems of low efficiency and insufficient accuracy currently faced in particle analysis processes, improving detection accuracy and efficiency, and bringing greater economic benefits.

Claims

1. A method for analyzing particles in cigarette filters, characterized in that, include: Data information of cigarettes was collected from different circumferential angles using CT equipment, and a three-dimensional reconstruction model of the cigarettes was constructed based on the data information; The granular regions are split from the 3D reconstruction model according to grayscale differences to form a point cloud dataset of the granular regions. Perform sphere fitting on the point cloud dataset of the granular region to obtain the center of each fitted sphere; For each fitted sphere, the distance from all its interior points to the center of the sphere is calculated, and the number of interior points whose distances are different from their corresponding radii is accumulated to obtain the number of abnormal interior points. When the number of abnormal interior points of a fitted sphere reaches the preset threshold for the number of interior points used to determine whether the particle surface is smooth, it is determined that the particle surface corresponding to the fitted sphere is not smooth.

2. The method for analyzing particles in a cigarette filter according to claim 1, characterized in that, The open3D open-source library was also used to process the point cloud data contained in each fitted sphere to obtain the corresponding surface area and volume; based on the surface area and volume of each fitted sphere, the sphericity index of the fitted sphere was calculated. The triangular mesh model algorithm is used to calculate the surface area, and the voxel mesh algorithm is used to calculate the volume.

3. The method for analyzing particles in a cigarette filter according to claim 2, characterized in that, The sphericity index is calculated using the following formula: ； Where V is the volume of the particle; S is the surface area of ​​the particle.

4. The method for analyzing particles in a cigarette filter according to claim 1, characterized in that, The radius of each fitted sphere is also calculated to obtain the average diameter of the particles.

5. The method for analyzing particles in a cigarette filter according to any one of claims 1-4, characterized in that, Perform the following steps on the point cloud dataset of the granular region until the number of interior points of all fitted spheres equals the number of interior points of the point cloud dataset of the granular region, thus obtaining all fitted spheres of the granular region: Select a preset number of data points from the point cloud dataset of the granular region, perform sphere fitting on the selected data points, and obtain the sphere corresponding to the selected data points when the fitting parameters of the fitted sphere are greater than the preset fitting parameter threshold.

6. The method for analyzing particles in a cigarette filter according to any one of claims 1-4, characterized in that, The granular regions are obtained from the 3D reconstructed model based on grayscale differences using the following method: Based on the data of the cigarette, a training model is generated to segment the particle regions according to the grayscale differences, and a training model for extracting particle regions is generated; the training model is called on the 3D reconstruction model to segment and obtain the particle regions.

7. The method for analyzing particles in a cigarette filter according to any one of claims 1-4, characterized in that, The data information of the cigarette is processed using a filtered back projection algorithm to generate a three-dimensional reconstruction model of the cigarette.

8. An analytical device for particles in cigarette filters, characterized in that, The device includes a processor for performing the analytical method for particles in a cigarette filter as described in any one of claims 1-7.

Images