Cigarette end face physical defect automatic detection method and system and storage medium

Abstract

The invention relates to the technical field of tobacco product quality detection, in particular to a cigarette end face physical defect automatic detection method and system and a storage medium, and the method comprises the steps: obtaining the three-dimensional point cloud data of the end face of a to-be-detected cigarette; obtaining a cigarette end face reference surface according to the three-dimensional point cloud data; geometric parameters of the cigarette end face are calculated according to the cigarette end face datum plane; and obtaining a cigarette end surface physical defect detection result according to the geometric parameters of the cigarette end surface. According to the embodiment of the invention, the complete three-dimensional point cloud data is obtained through one-time scanning, the end face reference plane is accurately extracted by using the RANSAC algorithm, and multi-index comprehensive evaluation is carried out based on the end face reference plane, so that wide coverage and comprehensive quantitative evaluation of various defects such as loose heads, contacts and spinning are realized.

Description

Technical Field

[0001] This invention relates to the field of tobacco product quality testing technology, specifically to an automatic detection method, system, and storage medium for physical defects on the end face of a cigarette stick. Background Technology

[0002] Detection of physical defects on cigarette end faces is a crucial step in cigarette production quality control, directly impacting product appearance and consumer experience. The current national standard for cigarettes, GB5606.3-2005, "Cigarettes Part III: Packaging, Rolling Technical Requirements and Storage and Transportation," clearly defines cigarette end face defects, primarily including typical defects such as empty ends, contact points, and unfinished cigarettes, and specifies concrete quantitative indicators. These defects not only affect the cigarette's appearance but also, to some extent, reflect the stability and consistency of the production process.

[0003] Currently, the detection of cigarette end-face defects on cigarette production lines mainly relies on manual visual inspection and measurement methods specified in GB / T22838.9-2008 "Determination of Physical Properties of Cigarettes and Filter Rods - Part 9: Empty Cigarette Ends" and GB / T22838.12-2008 "Part 12: Appearance of Cigarettes". This method suffers from limitations such as strong subjectivity, susceptibility to misjudgments and omissions, difficulty in adapting to the pace of modern large-scale production, and inability to achieve precise quantification and digital traceability of defect parameters. With the rapid development of digital manufacturing, intelligent quality inspection, and machine vision technologies, 3D imaging technology has been widely used in industrial defect detection, enabling high-precision 3D reconstruction of the surface morphology of minute objects. However, there is currently no structured light 3D imaging detection method specifically for cigarette end-face defects, and related systematic metrological traceability technologies are also lacking.

[0004] Therefore, there is an urgent need to establish a high-precision, non-contact, and automated method and system for detecting physical defects on cigarette end faces, so as to achieve accurate quantitative evaluation of three-dimensional defects on cigarette end faces, improve the detection efficiency and quality control level of the production line, and meet the needs of digital transformation in the industry. Summary of the Invention

[0005] The purpose of this invention is to provide an automatic detection method, system, and storage medium for physical defects on the end face of cigarettes, in order to solve the technical problems of low efficiency, high subjectivity, and high false positive and false negative rates caused by the existing technology that mainly relies on manual visual inspection and manual contact measurement for the detection of defects on the end face of cigarettes.

[0006] To achieve the above objectives, embodiments of the present invention provide an automatic detection method for physical defects on the end face of a cigarette stick, comprising: Acquire three-dimensional point cloud data of the end face of the cigarette stick to be tested; The reference plane of the cigarette end face is obtained based on the three-dimensional point cloud data; Calculate the geometric parameters of the cigarette end face based on the reference plane of the cigarette end face; The physical defect detection results of the cigarette end face are obtained based on the geometric parameters of the cigarette end face.

[0007] Optionally, obtaining the reference plane of the cigarette end face based on the three-dimensional point cloud data includes: The three-dimensional point cloud data is preprocessed, including voxel raster simplification and statistical filtering. Based on the preprocessed 3D point cloud data, RANSAC was used to extract the reference plane of the cigarette end face.

[0008] Optionally, based on the preprocessed 3D point cloud data, the reference plane of the cigarette end face is extracted using RANSAC, including: Three non-collinear points are randomly selected from the preprocessed 3D point cloud data; A planar model is built based on three non-collinear points extracted; Calculate the distance from each point in the preprocessed 3D point cloud data to the planar model; Each point is classified according to the calculated distance to obtain the set of interior points and the number of interior points in the set of interior points; Determine whether the iteration is complete; If the iteration is not completed, return to the step of randomly selecting three non-collinear points from the preprocessed 3D point cloud data; If the iteration is completed, the plane model corresponding to the maximum number of internal points is selected as the reference plane for the cigarette end face.

[0009] Optionally, calculating the geometric parameters of the cigarette end face based on the reference plane of the cigarette end face includes: Extract point cloud data of tobacco shreds inside the end face of the cigarette stick; The cavitation depth of the tobacco shreds is calculated according to formula (1). (1) in, The depth of the cavitation. The height of the reference plane at the end face of the cigarette stick. For the first The depth of the cavity filled with tobacco inside the block This represents the number of void defects.

[0010] Optionally, calculating the geometric parameters of the cigarette end face based on the reference plane of the cigarette end face includes: Fit the two-dimensional projection of the outer ring of the cigarette wrapping paper to obtain the total area of ​​the cigarette end face; Extract the three-dimensional point cloud data at a preset depth below the reference plane of the cigarette end face, and project it onto the XOY plane to obtain a two-dimensional image; Calculate the void section ratio according to formula (2). (2) in, The ratio of the void section, For the second two-dimensional image The area of ​​the vacant region, This represents the number of vacant areas. The total area of ​​the cigarette end face.

[0011] Optionally, calculating the geometric parameters of the cigarette end face based on the reference plane of the cigarette end face includes: The point cloud of the cigarette end face is sliced ​​along the Z-axis at a preset height step to obtain multiple slice planes parallel to the reference plane; The point cloud data of each slice plane is projected onto the corresponding two-dimensional raster image to obtain the void region; For each of the aforementioned void regions, extract its convex hull profile on the XOY plane and calculate the cross-sectional area enclosed by the convex hull profile. Calculate the void volume according to formula (3). (3) in, For the volume of the void, This represents the number of slice layers. For the first The height of the slice, For the first The cross-sectional area enclosed by the convex hull profile of the layer. This is the preset height step size.

[0012] Optionally, calculating the geometric parameters of the cigarette end face based on the reference plane of the cigarette end face includes: Based on the three-dimensional point cloud data, the outermost circumferential boundary of the cigarette end face is obtained by fitting. Detection is performed along the circumferential boundary direction to identify continuous point cloud segments that are lower than the reference plane by more than a preset threshold in the height direction, and these segments are determined to be contact areas; In the point cloud data of each contact area, the point with the smallest Z coordinate is selected as the lowest contact point; Calculate the contact depth according to formula (4). (4) in, For contact depth, , , , For the plane parameters of the reference plane, This is the lowest contact point; The three-dimensional point cloud data is projected onto the XOY plane to obtain a two-dimensional image of the cigarette end face. Extract the contact area from the two-dimensional image of the cigarette end face; Calculate the ratio of the total contact length to the circumference using formula (5). (5) in, The ratio of the total length of the contact to the circumference. For the first The central angle radians corresponding to each contact. This refers to the number of contacts.

[0013] Optionally, calculating the geometric parameters of the cigarette end face based on the reference plane of the cigarette end face includes: Extract point cloud data higher than the reference plane of the cigarette end face from the point cloud data of the cigarette end face, and use it as the point cloud data for cigarette spinning; Obtain the maximum value of the point cloud data in the Z-axis direction; Calculate the silk-spinning length according to formula (6). (6) in, The length of the silk thread. This represents the maximum value of the point cloud data along the Z-axis. The height of the reference plane at the end face of the cigarette stick; Calculate the average height of the silk-spinning point cloud data along the Z-axis; Calculate the end face flatness according to formula (7). (7) in, For end face flatness, The first point cloud data for spinning silk The height value of each point. The average height of the point cloud data for spinning silk. This represents the total number of points in the point cloud data.

[0014] On the other hand, the present invention also provides an automatic detection system for physical defects on the end face of a cigarette stick, the system comprising: A line structured light 3D imaging module is used to acquire 3D point cloud data of the end face of the cigarette stick to be tested; A precision moving platform is used to support and fix the cigarette to be inspected, so that the end face of the cigarette is kept within the appropriate measurement range of the line structured light 3D imaging device; The control module is used to control the precision moving platform to move along a set trajectory; The data processing module is used to obtain the reference plane of the cigarette end face based on the three-dimensional point cloud data, calculate the geometric parameters of the cigarette end face, and output the detection results of physical defects of the cigarette end face. A processor configured to perform any of the methods described above.

[0015] In another aspect, the present invention also provides a computer-readable storage medium storing instructions that, when executed by a processor, implement any of the methods described above.

[0016] This invention applies line structured optical triangulation technology to the detection of physical defects on the cigarette end face. Combined with high-precision point cloud processing and intelligent algorithms, it enables automated identification and quantitative analysis of various defects on the cigarette end face, such as hollow ends, contact points, and cigarette fabrication. Specifically: The present invention utilizes voxel grid simplification and statistical filtering to effectively remove noise point clouds, achieving high-resolution reconstruction of minute details on the cigarette end face. The RANSAC algorithm is used to accurately extract the end face reference plane, avoiding the problems of traditional plane fitting being easily affected by noise and local anomalies, thereby improving the measurement accuracy of cavity depth, contact depth, and cigarette tip height.

[0017] The embodiments of the present invention construct an automated scanning system that links a precision moving platform with a line laser and a camera, and integrate a data processing module that can automatically perform three-dimensional reconstruction and defect calculation. This enables continuous and rapid inspection on the production line without the need for repeated manual sampling and measurement. The inspection of the end faces of multiple cigarettes can be completed in a very short time, thereby greatly improving the inspection efficiency and automation level.

[0018] The embodiments of this invention acquire complete three-dimensional point cloud data in a single scan, and then conduct a comprehensive evaluation based on multiple indicators (such as the depth, volume, and cross-sectional ratio of the hollow part, the circumferential ratio and depth of the contact, the maximum length of the spinneret, and the flatness of the end face). This achieves broad coverage and comprehensive quantitative evaluation of various defects such as hollow parts, contacts, and spinnerets. At the same time, it provides a reliable data foundation and general method support for quality traceability, metrological calibration, and extended applications to cigarettes, filter rods, and other products of different specifications.

[0019] Other features and advantages of the embodiments of the present invention will be described in detail in the following detailed description section. Attached Figure Description

[0020] The accompanying drawings are provided to further illustrate embodiments of the present invention and form part of the specification. They are used together with the following detailed description to explain the embodiments of the present invention, but do not constitute a limitation thereof. In the drawings: Figure 1 A flowchart of an automatic detection method for physical defects on the end face of a cigarette according to an embodiment of the present invention; Figure 2 A flowchart of a method for voxel raster simplification of three-dimensional point cloud data according to an embodiment of the present invention; Figure 3 A flowchart of a method for statistical filtering of three-dimensional point cloud data according to an embodiment of the present invention; Figure 4 This is a schematic diagram of noise points in a statistical filtering process according to an embodiment of the present invention; Figure 5 A flowchart of a method for extracting a reference plane at the end face of a cigarette according to an embodiment of the present invention; Figure 6 This is a schematic diagram of the precision moving platform structure of an automatic detection system for physical defects on the end face of a cigarette according to an embodiment of the present invention. Figure 7 This is a reconstruction diagram of the curved surface of a cigarette end face according to an embodiment of the present invention; Figure 8 This is a schematic diagram illustrating the calculation of short position parameters according to an embodiment of the present invention; Figure 9 This is a schematic diagram illustrating the calculation of contact parameters according to an embodiment of the present invention; Figure 10 This is a schematic diagram illustrating the calculation of silk-spinning parameters according to an embodiment of the present invention; Figure 11 This is a schematic diagram of the measurement results of the air head parameter according to one embodiment of the present invention; Figure 12 This is a schematic diagram of contact parameter measurement results according to one embodiment of the present invention; Figure 13 This is a schematic diagram of the measurement results of the silk-spinning parameters according to one embodiment of the present invention. Detailed Implementation

[0021] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustration and explanation only and are not intended to limit the scope of the present invention.

[0022] It should be noted that the acquisition, transmission, storage, use, and processing of data in the technical solution of this application all comply with relevant laws and regulations. In the embodiments of this application, certain existing industry solutions such as software, components, and models may be mentioned. These should be considered exemplary, intended only to illustrate the feasibility of implementing the technical solution of this application, and do not imply that the applicant has already used or necessarily used such solutions.

[0023] like Figure 1 The diagram shows a flowchart of an automatic detection method for physical defects on the end face of a cigarette stick according to an embodiment of the present invention. Figure 1 The detection method may include the following steps: In step S10, the three-dimensional point cloud data of the end face of the cigarette to be tested is acquired; In step S11, the reference plane of the cigarette end face is obtained based on the three-dimensional point cloud data; In step S12, the geometric parameters of the cigarette end face are calculated based on the reference plane of the cigarette end face; In step S13, the physical defect detection results of the cigarette end face are obtained based on the geometric parameters of the cigarette end face.

[0024] In such Figure 1 In the automatic detection method for physical defects on the cigarette end face shown, step S10 is used to acquire three-dimensional point cloud data of the cigarette end face to be tested. In this embodiment, a high-precision line laser camera three-dimensional scanning device can be used to scan the cigarette end face, and the surface data of the cigarette end face is collected line by line, and the scan data of each line is stitched together to form complete point cloud data.

[0025] Step S11 is used to obtain the reference plane of the cigarette end face based on the three-dimensional point cloud data. In this embodiment, the specific method for obtaining the reference plane of the cigarette end face in step S11 can be of various forms known to those skilled in the art. In one example of the present invention, step S11 may include: In step S20, the 3D point cloud data is preprocessed, including voxel raster simplification and statistical filtering. In step S21, based on the preprocessed three-dimensional point cloud data, the reference plane of the cigarette end face is extracted using RANSAC.

[0026] Step S20 is used to preprocess the 3D point cloud data, including voxel grid simplification and statistical filtering. Specifically, voxel grid simplification involves setting the voxel unit size, such as 0.05mm to 0.1mm, according to the size and accuracy requirements of the cigarette end face, to further reduce the volume of the original point cloud. Then, all points within each voxel are replaced with centroid points, preserving geometric features while improving the efficiency of subsequent processing. Specifically, in this example, the specific method of voxel grid simplification can be of various forms known to those skilled in the art. In one example of the present invention, the specific method of voxel grid simplification can be as follows: Figure 2 The steps shown are described. Figure 2 In this context, methods for voxel raster simplification can include: In step S30, the maximum and minimum coordinate values ​​in the x, y, and z directions are obtained based on the original point cloud data. In step S31, the grid side length is set according to the maximum and minimum coordinate values, and voxels are divided; In step S32, the grid number of each voxel is obtained.

[0027] In step S33, empty voxels in the voxel grid are deleted. An empty voxel is a voxel that does not contain any point cloud data. In step S34, for each non-empty voxel, the centroid of all point cloud coordinates inside it is calculated. In step S35, the calculated centroid point is used as the unique representative point of the voxel, and all other point cloud data in the voxel except for the centroid point are deleted.

[0028] In such Figure 2 In the method shown, step S30 is used to determine the maximum value of the three coordinate axes based on the input point cloud data. and minimum value Step S31 is used to determine the side length of the minimum bounding box based on the distance between the maximum and minimum values ​​within the range of the three coordinate axes. The side length of each small grid is set to L. The number of small grids into which the side length of the minimum bounding box of the three coordinate axes is evenly divided is M, N, and O respectively. In order to prevent point cloud data from appearing at the boundary of the bounding box, the number of grids is increased by 1. The voxels are divided according to formula (8): (8) in, for The number of grids divided in the direction, for The number of grids divided in the direction, for The number of grids divided in the direction, The grid side length , , They represent all points in axis, axis, Maximum coordinate value in the axis direction, , , They represent all points in axis, axis, Minimum coordinate value along the axis. This indicates rounding down to the nearest integer.

[0029] Step S32 is used for any point in the point cloud Calculate the index number along the length, width, and height. Calculate the raster number to which the point belongs based on the index number. Specifically, in this example, the raster number of each voxel can be obtained, for example, according to formulas (9) and (10): (9) (10) in, Represents any point in a point cloud Along the index numbers of length, width, and height, For grid numbering.

[0030] Steps S33 to S35 are used to delete rasters without point cloud data, recalculate the centroid of all points with the same number, and replace other points with the centroid. This method can reduce the number of point clouds while preserving the geometric features of the point cloud and improve processing speed.

[0031] After simplifying the point cloud data in step S20, statistical filtering is required to remove noise points and eliminate outliers caused by environmental interference or abnormal local reflections. In this example, the specific method of statistical filtering can be of various forms known to those skilled in the art. In one example of the present invention, the specific method of statistical filtering can be as follows: Figure 3 The steps shown are described. Figure 3 In statistical filtering, methods can include: In step S40, a KD-Tree topology is constructed from the voxel raster-simplified point cloud data; In step S41, the average distance from any point to its k nearest neighbors is calculated; In step S42, the Gaussian distribution of the point cloud data is obtained and a threshold range is set; In step S43, the point to be filtered is obtained as the current point; In step S44, it is determined whether the average distance of the current point is within a set threshold range; In step S45, if the average distance of the current point is not within the set threshold range, the current point is removed. In step S46, it is determined whether there are still points to be filtered; In step S47, if it is determined that there are still points to be filtered, the process returns to the step of obtaining the points to be filtered as the current point.

[0032] In such Figure 3 In the method shown, step S40 is used to construct a KD-Tree topology structure from the voxel raster-simplified point cloud data. By establishing an efficient spatial index, the complexity of nearest neighbor search is greatly reduced. Step S41 is used to arbitrarily select a point in the point cloud. Find the k nearest neighbors of the point (including itself), and calculate the value of any point according to formulas (11) and (12). The average distance to the k nearest neighbors. (11) (12) in, For any point The average distance to the k nearest neighbors. for Point to nearest neighbor The distance.

[0033] Step S42 is used to obtain the Gaussian distribution of the point cloud data and set a threshold range. Specifically, in this example, a certain point in the point cloud data... Average distance to other points in its neighborhood Satisfies a Gaussian distribution: (13) in, Average distance to all points The mean, Average distance to all points The standard deviation and threshold interval are In this example, such as Figure 4 As shown, we can assume the number of nearest neighbors is 5, and the red circle represents the standard threshold. The black circle represents the average distance from the nearest neighbor to the given local area. It can be seen that the average distance from pa to its nearest neighbor is within the threshold, therefore this point is not isolated. However, the average distance from pb to its nearest neighbor is outside the threshold range, therefore this point is isolated and needs to be removed.

[0034] Steps S43 to S47 are used to traverse each point and determine the value of that point. Whether it is within the threshold range, if If the value is within the threshold range, then retain the point; if... If the value is not within the threshold range, it is considered noise and removed.

[0035] Step S21 is used to extract the reference plane of the cigarette end face based on the preprocessed 3D point cloud data using RANSAC. In this embodiment, the specific method for extracting the reference plane of the cigarette end face in step S21 can be of various forms known to those skilled in the art. In one example of the present invention, step S21 may include, for example... Figure 5 The steps shown are described. Figure 5 In this context, step S21 may include: In step S50, three non-collinear points are randomly selected from the preprocessed 3D point cloud data; In step S51, a planar model is established based on the three non-collinear points extracted. In step S52, the distance from each point in the preprocessed 3D point cloud data to the planar model is calculated; In step S53, each point is classified according to the calculated distance to obtain the set of interior points and the number of interior points in the set of interior points; In step S54, it is determined whether the iteration has been completed; In step S55, if it is determined that the iteration has not been completed, the process returns to the step of randomly selecting three non-collinear points from the preprocessed 3D point cloud data. If the iteration is completed, the plane model corresponding to the maximum number of internal points is selected as the reference plane for the cigarette end face.

[0036] In such Figure 5 In the method shown, step S50 is used for random sampling, randomly selecting three non-collinear points from the filtered point cloud data. , , The requirement that the three points are not collinear ensures that a unique plane can be fitted; if the three points are collinear, a new plane is selected.

[0037] Step S51 is used to build a planar model based on the three extracted non-collinear points. Specifically, in this example, it can be done by first using the coordinates of the three points... , , The vector can be constructed, for example, according to formulas (14) and (15). (14) (15) in, and For The vector constructed as a reference point , , These are three non-collinear points.

[0038] Through vector cross product The plane normal vector can be obtained, for example, by using formula (16). : (16) Will Substitute the parameters of the plane equation into formula (17) to solve for the parameters of the plane equation. , (17) in, , , , represents the parameters of the plane equation.

[0039] In step S52, the distance from each point in the filtered point cloud data to the planar model is calculated. Specifically, in this example, for any point in the point cloud P... Based on the obtained plane equation (A,B,C,D), the distance from the point to the plane is calculated using formula (18). (18) in, For any point Distance to the planar model, , , , represents the parameters of the plane equation.

[0040] Step S53 is used to classify each point according to the calculated distance to obtain the set of inliers and the number of inliers in the set. Specifically, in this example, the specific method for obtaining the set of inliers and the number of inliers in the set may include: In step S60, any unclassified point is selected as the current point; In step S61, it is determined whether the distance from the current point to the current plane model is less than or equal to a threshold. In step S62, if the distance from the current point to the current plane model is less than or equal to a threshold, the current point is added to the set of interior points. In step S63, it is determined whether there are any unclassified points; In step S64, if it is determined that there are unclassified points, return to the step of selecting any unclassified point as the current point; If no unclassified points are found, count the number of interior points in the current model.

[0041] Steps S60 to S64 are used to compare the distances from all points to the plane with a threshold. If If a point is found to be an interior point of the plane, then that point is considered an "inlier". All points whose distance is less than or equal to a threshold are considered interior points. The points are included in the interior point set I, and the number of interior points in the current model |I| is counted.

[0042] Steps S54 and S55 are used to repeat the process of random sampling, model estimation, and interior point statistics, trying different combinations of the three points in each iteration. When the number of iterations reaches k... max Alternatively, if a sufficiently high-quality model has been found (e.g., the proportion of interior points exceeds a certain threshold), the RANSAC process ends. Finally, the optimal model M is selected. best The interior point set I of the optimal model (i.e., the set with the most interior points, the smallest total residual, or a combination of both) can be used as the reference plane for the end face. In this example, the interior point set I of the optimal model can also be used. bestPerform least-squares plane fitting again to further improve accuracy and obtain more accurate plane parameters. If necessary, mean or regression can be performed on the set of interior points to make the plane parameters more evenly distributed across all interior points.

[0043] Step S12 is used to calculate the geometric parameters of the cigarette end face based on the reference plane of the cigarette end face. In this embodiment, the geometric parameters of the cigarette end face to be calculated may include parameters for void defects, contact defects, and wire-spinning defects. The void defect parameters may include void depth, void cross-sectional ratio, and void volume. In this example, the method for calculating the void depth may include the following steps: In step S70, extract the tobacco point cloud data inside the end face of the cigarette stick; In step S71, the cavitation depth of the tobacco shreds is calculated according to formula (1). (1) in, The depth of the cavitation. The height of the reference plane at the end face of the cigarette stick. For the first The depth of the cavity filled with tobacco inside the block This represents the number of void defects.

[0044] Step S70 obtains a reference plane by fitting the point cloud data of the outer ring cigarette paper, calculates the height value of the reference plane in the Z-axis direction, and extracts the point cloud data of tobacco shreds inside the cigarette end face. Step S71 is used to obtain the cavity depth based on the height of the reference plane and the average height of the tobacco shreds in the region.

[0045] In this example, the calculation method for the void section ratio may include the following steps: In step S80, the two-dimensional projection of the outer ring of the cigarette wrapping paper is fitted to obtain the total area of ​​the cigarette end face; In step S81, three-dimensional point cloud data at a preset depth below the reference plane of the cigarette end face is extracted and projected onto the XOY plane to obtain a two-dimensional image; In step S82, the void section ratio is calculated according to formula (2). (2) in, The ratio of the void section, For the second two-dimensional image The area of ​​the vacant region, This represents the number of vacant areas. This represents the total area of ​​the cigarette end face.

[0046] Step S80 involves fitting a two-dimensional circle based on the two-dimensional projection of the outer packaging paper. The area of ​​this circle represents the total area of ​​the cigarette end face. Step S81 involves extracting a three-dimensional point cloud at a depth h below the reference plane within the projected circular area and projecting this point cloud onto the XOY plane to generate a two-dimensional image. In this example, the value of h can be 1 mm. In this image, the hollow areas are marked in black. Step S82 involves calculating the ratio of the total area of ​​these black areas to the total area of ​​the end face, thus obtaining the hollow section ratio. In this example, the method for calculating the void volume may include the following steps: In step S90, the point cloud of the cigarette end face is sliced ​​along the Z-axis at a preset height step to obtain multiple slice planes parallel to the reference plane. In step S91, the point cloud data of each slice plane is projected onto the corresponding two-dimensional raster image to obtain the void area; In step S92, for each layer of the void region, the convex hull profile on the XOY plane is extracted, and the cross-sectional area enclosed by the convex hull profile is calculated. In step S93, the vacancy volume is calculated according to formula (3). (3) in, For the volume of the void, This represents the number of slice layers. For the first The height of the slice, For the first The cross-sectional area enclosed by the convex hull profile of the layer. This is the preset height step size.

[0047] Step S90 involves slicing the end-face point cloud into layers along the Z-axis and accumulating the volume based on the convex hull area of ​​the black regions in each layer. Specifically, parallel slices are constructed layer by layer between the end-face point cloud and the reference plane according to a preset height step, and slicing is terminated when the slice height is less than 1 mm from the reference plane. Step S91 projects the point cloud data of each slice onto a two-dimensional raster image and sets the raster cells not occupied by the point cloud to black to represent the spatial void area at that height. Step S92 extracts the convex hull contour of each black region in the XOY plane and calculates the cross-sectional area corresponding to the convex hull. Step S93 discretizes and sums the convex hull areas of all layers according to the height interval of each slice to obtain the total volume of the void. V .

[0048] In step S12, the contact defect parameters can include contact depth and the contact length as a percentage of the circumference. To obtain the contact area, the circumferential boundary can first be obtained by fitting the outermost circle of the end face, and then continuous point cloud segments below a certain threshold in the height direction of the reference plane can be detected along the circumferential direction to calculate the contact depth and the contact length as a percentage of the circumference. Specifically, in this example, the method for calculating the contact depth can include the following steps: In step S100, the outermost circumferential boundary of the cigarette end face is obtained by fitting based on the three-dimensional point cloud data; In step S101, detection is performed along the circumferential boundary direction to identify continuous point cloud segments that are lower than the reference plane by more than a preset threshold in the height direction, and the segments are determined as contact areas. In step S102, the point with the smallest Z coordinate in the point cloud data of each contact area is selected as the lowest contact point; In step S103, the contact depth is calculated according to formula (4). (4) in, For contact depth, , , , For the plane parameters of the reference plane, This is the lowest contact point.

[0049] Steps S100 to S101 are used to obtain the contact area, step S102 is used to find the lowest contact point of all contacts, and then step S103 is used to calculate the contact point depth.

[0050] In this example, the method for calculating the contact length as a percentage of the circumference can include the following steps: In step S110, the three-dimensional point cloud data is projected onto the XOY plane to obtain a two-dimensional image of the cigarette end face; In step S111, the contact area is extracted from the two-dimensional image of the cigarette end face; In step S112, the proportion of the total contact length to the circumference is calculated according to formula (5). (5) in, The ratio of the total length of the contact to the circumference. For the first The central angle radians corresponding to each contact. This refers to the number of contacts.

[0051] Steps S110 to S112 involve projecting the cigarette end face onto the XOY plane to obtain a two-dimensional image of the cigarette end face, extracting the contact area based on the two-dimensional image, and finding the endpoint of each contact. By calculating the curvature of each contact, the circumference ratio occupied by the contact is obtained.

[0052] The parameters for the wire-spinning defect in step S12 can include wire-spinning length and end-face flatness. In this example, the calculation method for the wire-spinning length can include the following steps: In step S120, point cloud data higher than the reference plane of the cigarette end face is extracted from the point cloud data of the cigarette end face and used as the point cloud data of the cigarette spinning. In step S121, the maximum value of the silk-spinning point cloud data in the Z-axis direction is obtained; In step S122, the silk-spinning length is calculated according to formula (6). (6) in, The length of the silk thread. This represents the maximum value of the point cloud data along the Z-axis. This is the height of the reference plane at the end of the cigarette stick.

[0053] Steps S120 to S122 are used to extract the end face point cloud region that is higher than a certain threshold above the reference plane, find the vertical distance between the highest point in the region and the reference plane, and use it as the silk-spinning length.

[0054] In this example, the method for calculating the end face flatness may include the following steps: In step S130, the average height of the silk-spinning point cloud data in the Z-axis direction is calculated; In step S131, the end face flatness is calculated according to formula (7). (7) in, For end face flatness, The first point cloud data for spinning silk The height value of each point. The average height of the point cloud data for spinning silk. This represents the total number of points in the point cloud data.

[0055] Steps S130 to S131 are used to extract the tobacco shred region based on the point cloud of the cigarette end face and calculate its average height in the Z-axis direction. Using the reference height as a baseline, the variance of the height deviation of the end face point cloud in the Z-axis direction is calculated to obtain a quantitative value for characterizing the flatness of the cigarette end face.

[0056] Step S13 is used to obtain the physical defect detection results of the cigarette end face based on the geometric parameters of the cigarette end face. After extracting the defect areas of hollow end, contact end, and silk ejection, the corresponding parameters are compared with preset thresholds. If the threshold is exceeded, it is judged as a defect and classified and recorded.

[0057] On the other hand, the present invention also provides an automatic detection system for physical defects on the end face of a cigarette stick, the system comprising: a line structured light three-dimensional imaging module, a precision moving platform, a control module, a data processing module, and a processor.

[0058] The line structured light 3D imaging module is used to acquire 3D point cloud data of the cigarette end face under test. The line structured light 3D imaging module includes a line laser emitter and a camera imaging module. The line laser emitter and camera imaging device work together; in this embodiment, a vertically incident optical triangulation structure is used to reduce measurement errors caused by the tilted projection of the laser. The camera imaging module can use a CMOS image sensor with a tilt-shift lens to capture the reflected stripes of the laser line on the cigarette end face, thereby acquiring high-resolution image data. The camera is connected to the data processing unit via a data cable.

[0059] A precision moving platform is used to support and fix the cigarette to be inspected, keeping the cigarette end face within the appropriate measurement range of the line structured light 3D imaging device. Precise and controllable reciprocating movement is achieved through a stepper motor and photoelectric limit switches, ensuring uniform movement in the X / Y / Z axis directions, or at least in the Y axis direction, scanning the cigarette end face row by row or segment by segment. Figure 6 As shown, the Z-axis of the system consists of a cylindrical slide bar, with the line laser sensor fixed on the slide rail and its height manually adjustable. Due to the limited Z-axis measurement range of the line laser sensor, the distance between the line laser sensor and the object being measured needs to be adjusted by moving the Z-axis slide bar to ensure the object is within the sensor's measurement range. The X-axis represents the width of the line laser scan, which is related to the Z-axis distance between the line laser and the end face of the cigarette being measured. The Y-axis represents the direction of slide table movement, moving with a pulse equivalent of 0.01 mm to scan the measured contour. To prevent the moving platform from exceeding its travel range, photoelectric limit switches are installed at both ends of the slide table.

[0060] The control module is used to control the movement of the precision moving platform along a set trajectory. It consists of a stepper motor driver and a motion controller, which are responsible for receiving external commands and driving the precision moving platform to move along the set trajectory; photoelectric limit switches limit the movement range, and trigger limit signals when the slide reaches the set travel limit to prevent hardware damage caused by misoperation.

[0061] The data processing module is used to obtain the reference plane of the cigarette end face based on 3D point cloud data, calculate the geometric parameters of the cigarette end face, and output the detection results of physical defects in the cigarette end face. In this example, the data processing module can be an industrial computer or an embedded processor, used to perform laser center extraction, point cloud reconstruction, and defect identification on the image frames transmitted by the camera. The built-in software algorithms include camera calibration, laser line calibration, voxel grid simplification, statistical filtering, RANSAC plane extraction, and calculation functions for defect parameters such as voids, contacts, and wire ejection. The results are output to the display and alarm module or other production process control systems. The processor is configured to execute any of the methods described in the automatic detection method for physical defects in the cigarette end face.

[0062] The aforementioned automatic detection system for physical defects on the cigarette end face may also include a display and alarm module for real-time presentation of the detection process and results. When a detected defect exceeds a preset threshold, an alarm signal is issued or a control command is sent to the host computer so that on-site operators can handle the situation promptly.

[0063] In another aspect, the present invention also provides a computer-readable storage medium storing instructions that, when executed by a processor, implement any of the methods described above.

[0064] This invention also includes an experiment on the end face inspection of cigarettes. By conducting detailed inspections of the end faces of cigarettes of different specifications available on the market, the proportion and severity of hollow ends, contact points, and filament extrusion were analyzed, providing data support for optimizing cigarette manufacturing processes. The experimental design considered different cigarette specifications to ensure the universality and representativeness of the test results. Details are as follows: In order to better analyze the quality defects of cigarette end faces, the experiment divided cigarettes into three categories according to their diameter: (1) Coarse cigarettes: These are cigarettes with a larger diameter. These cigarettes have a higher tobacco filling density and the end faces are usually more regular. (2) Medium cigarettes: These are between coarse and thin cigarettes and represent the more common cigarette specifications on the market. (3) Thin cigarettes: These are cigarettes with a smaller diameter. Due to the more compact space, these cigarettes are prone to uneven filling at the end faces.

[0065] First, sample preparation is conducted. A certain number of samples are selected for each type of cigarette to ensure that the samples are diverse and representative. The specific sample size can be determined based on the cigarette market share and production batch details. Second, data acquisition is performed. A cigarette end-face inspection platform is used to perform high-precision scanning of slim, medium, and thick cigarettes. Surface data of the cigarette end face is collected line by line, and the scan data from each line is stitched together to form a complete point cloud. Next, the data is processed. The obtained point cloud data is filtered to remove noise and outliers, obtaining more accurate 3D information of the cigarette end face. Surface reconstruction technology is used to convert the point cloud data into a continuous 3D surface model, constructing a complete 3D structure of the cigarette end face. Based on this, such as... Figure 7 As shown, using visualization tools to display the 3D model can clearly present the geometric shape and details of the cigarette end face. Figure 7 Part (a) shows the reconstructed surface of the end face of a thick cigarette, part (b) shows the reconstructed surface of the end face of a medium cigarette, and part (c) shows the reconstructed surface of the end face of a thin cigarette. Finally, each type of cigarette was analyzed separately to obtain data on the average incidence and severity distribution of defects in the cigarette tip, contact point, and cigarette extrusion for the three types of cigarettes. By comparing the test results of different specifications of cigarettes, the degree of influence of cigarette tip defects on different specifications was determined.

[0066] The current national standard for cigarettes, GB5606.3-2005, "Cigarettes Part III: Packaging, Rolling Technical Requirements and Storage and Transportation," clearly stipulates the requirements for cigarette appearance, including: "empty ends (end cavity depth > 1.0 mm, cavity cross-section ratio > 2 / 3)" defects at the cigarette end face. Based on the above measurement standards, the empty end parameters of the three-dimensional point cloud data are calculated as follows: ① Determine the reference plane: such as... Figure 8 The height of the outer ring of packaging paper shown in section (a) is recorded as the reference plane height of 0, providing a standard for subsequent calculations. The reference plane needs to be accurately determined to ensure the accuracy of the calculations. ② Measurement of end cavity depth: such as Figure 8 As shown in section (b), the internal filling portion of the tobacco is extracted using radius filtering, and the outer packaging paper and other interfering data are removed. The extracted point cloud data is then analyzed to locate localized cavity regions and calculate their heights. Finally, the depth of the cavity inside the tobacco is determined by averaging the heights of these cavities. The distance between the calculated reference plane and the internal cavity depth yields the true cavity depth. ③ Cavity volume: as shown... Figure 8 As shown in section (b), the cut tobacco filling portion is sliced ​​into convex sections, and the volume of each convex section is calculated. Finally, the sum of the volumes of each section is used to obtain the void volume. ④ Void section ratio measurement: as shown in section (b). Figure 8As shown in section (c), a horizontal cut is made at the end of the cigarette, and the circular surface of the cigarette end face is fitted by the circumscribed circle and the area of ​​the cigarette end face is calculated; the area of ​​the black part inside the circular surface is statistically analyzed to obtain the void area, and finally the two are compared to calculate the void ratio.

[0067] According to the standard "Determination of Physical Properties of Cigarettes and Filter Rods Part 12: Cigarette Appearance", the specific steps for manual contact measurement are as follows: visually inspect the cigarette end face contact, mark both ends of the contact and the deepest point of the contact, cut open the cigarette, remove the tobacco and tobacco bundle, unfold the cigarette paper and tipping paper, and measure the contact length and cigarette circumference with a steel ruler; also measure the length from the maximum contact depth to the cut, and calculate the proportion of the length to the circumference. The three-dimensional point cloud contact parameters are calculated according to the standard measurement steps: ① Contact depth calculation: such as... Figure 9 As shown, the contact portion is cut out based on the outer circle fitted to the outer end face, and the lowest point of the contact is found using an extreme value algorithm. The contact depth is then obtained by subtracting it from the reference plane. ② The contact portion is unfolded in a plane, and the length of the contact portion is calculated. The ratio of the contact length to the circumference is calculated by comparing it with the circumference of the outer end face. Figure 9 As shown.

[0068] In cigarette production, empty ends and contact points are common defects. In addition, cigarette ends may also have appearance defects such as unfinished tobacco strands and inclusions. The standard "Determination of Physical Properties of Cigarettes and Filter Rods Part 12: Cigarette Appearance" specifies the detailed measurement procedures for manually determining the "contact point" and "exposed tobacco strands" on the cigarette end face. Based on the standard procedures, the unfinished tobacco strand parameters are calculated as follows: ① Unfinished tobacco strand length calculation: (e.g., ...) Figure 10 As shown, cut out internal tobacco shreds that are greater than the height of the cigarette end face, calculate the maximum value of the shredded tobacco, and subtract it from the reference height to obtain the length of the tobacco shreds. ② End face flatness calculation: as shown Figure 10 As shown, internal tobacco shreds that are greater than the height of the cigarette end face are cut out, and the height variance of the protruding part is calculated, i.e., the end face flatness.

[0069] For each of the three sample types, ten cigarettes were taken for statistical analysis to calculate the cavity depth, cavity volume, and cavity cross-sectional ratio. The measurement results are as follows: Figure 11 The diagram illustrates the trends in cavity depth, cavity volume, and cavity cross-sectional area ratio for slim, medium, and coarse cigarettes under different measurement cycles. Figure 11 As can be seen in section (a), the cavity depth of thicker cigarettes is significantly greater than that of thinner and medium-sized cigarettes, and the depth of thicker cigarettes fluctuates more significantly in multiple measurements, while the changes in thinner and medium-sized cigarettes are more stable. Figure 11 In section (b), the void volume of the coarse branch is significantly higher than that of the medium and fine branches, especially in the first few measurements where the volume fluctuates more, while the volume of the medium and fine branches is smaller and fluctuates less. Figure 11Section (c) shows the variation in the void cross-sectional area ratio. The void cross-sectional area ratio is generally higher in medium-sized cigarettes and fluctuates more significantly, while the variation is smaller in thin and thick cigarettes, with the thin cigarettes showing the most stable change. These data indicate that the void characteristics of cigarettes are displayed in multiple dimensions and exhibit different levels of volatility.

[0070] For each of the three sample types, ten cigarettes were taken for statistical analysis to calculate the contact depth and the contact tip circumference ratio. Multiple measurements of contact depth and contact tip circumference ratio for slim, medium, and coarse cigarettes are presented. Figure 12 Part (a) shows the contact depth variation of the three types of cigarettes, which ranges from -0.8 mm to 2.4 mm, with an overall value not exceeding 2 mm; Figure 12 Part (b) shows the contact circumference ratio, which ranges from -0.4 to 1.0, and generally does not exceed 1 / 3. Both figures illustrate that during the measurement process, the contact depth and contact circumference ratio of various cigarette types were within the specified range. The data shows that contact defects account for a small percentage of finished cigarettes; generally, cigarettes on the market rarely have contact problems.

[0071] For each of the three sample types, ten cigarettes were taken for statistical analysis, and the silk length and end-face flatness were calculated. The figure shows the changing trends of silk length and end-face flatness for slim, medium, and coarse cigarettes under different measurement cycles. Figure 13 In section (a), the length of the cigarette spit is relatively stable but generally low, while the length of the cigarette spit fluctuates more, especially showing a significant increase in the 4th and 8th measurements. The length of the cigarette spit is relatively stable and slightly lower than that of the cigarette spit. Figure 13 Part (b) shows the trend of end-face flatness. The end-face flatness of slim cigarettes fluctuated more in different measurements and was generally higher than that of medium and thick cigarettes, while the end-face flatness of medium and thick cigarettes was relatively lower and fluctuated less. This indicates that different types of cigarettes show significant differences in their output length and end-face flatness, and the measurement of output parameters is particularly important for controlling the overall quality of cigarettes.

[0072] This invention, by combining high-precision point cloud processing and intelligent algorithms, enables the automated identification and quantitative analysis of various defects on the cigarette end face, such as hollow ends, contacts, and cigarette burnout. Compared with existing methods relying on manual labor or two-dimensional image plane detection, this invention has the following beneficial technical effects: 1. High detection accuracy and accurate defect quantification: This invention adopts a vertical incident triangulation measurement structure to reduce measurement errors caused by tilted projection; at the same time, it effectively removes noise point clouds by using voxel grid simplification and statistical filtering to achieve high-resolution reconstruction of tiny details on the cigarette end face; furthermore, it uses the RANSAC algorithm to accurately extract the end face reference plane, avoiding the problem of traditional plane fitting being easily affected by noise and local abnormal data, thereby improving the measurement accuracy of cavitation depth, contact depth and spin height.

[0073] 2. High degree of automation and significantly improved inspection efficiency: The end face is scanned line by line through the linkage of a precision moving platform, line laser and camera. The data processing module can automatically perform three-dimensional reconstruction and defect calculation. There is no need for repeated manual sampling and measurement. It can be continuously and quickly inspected on the production line. The end face inspection of multiple cigarettes only takes a very short time, which is suitable for modern large-scale production.

[0074] 3. Wide coverage of defect types and diversified evaluation indicators: It can comprehensively evaluate three-dimensional indicators such as depth, volume and cross-sectional ratio for "empty head" defects, and can also quantify the circumferential ratio and depth of "contact" on the end face, the maximum length of the "spinning" area and the flatness of the end face; a complete three-dimensional point cloud can be obtained through a single scan, providing basic data for subsequent analysis or other quality indicator evaluation (such as end-face spin prediction).

[0075] 4. Excellent traceability and metrological calibration performance: The system can be calibrated with specially designed standard workpieces and uses point cloud registration method to monitor calibration parameters over a long period of time, ensuring that the test results have metrological traceability; the test results are consistent with industry or national metrological standards, providing a reliable basis for the digital control of cigarette appearance quality.

[0076] 5. Strong compatibility and scalability: This method is not only applicable to cigarettes of different diameters (such as slim, medium and thick cigarettes), but also to the end face inspection of filter rods or other similar columnar products; the point cloud preprocessing and RANSAC algorithm are universal and can be flexibly transferred to other 3D inspection systems to achieve more dimensional appearance inspection or industrial measurement.

[0077] Through the above innovative design and optimization, this invention achieves efficient and automated detection of physical defects on the cigarette end face. It can not only significantly improve the accuracy and speed of detection, but also provide quantitative parameter support for quality management and process improvement, and has broad industrial application value and promotion prospects.

[0078] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0079] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0080] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0081] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0082] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0083] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0084] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0085] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0086] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. An automatic detection method for physical defects on the end face of a cigarette stick, characterized in that, The detection method includes: Acquire three-dimensional point cloud data of the end face of the cigarette stick to be tested; The reference plane of the cigarette end face is obtained based on the three-dimensional point cloud data; Calculate the geometric parameters of the cigarette end face based on the reference plane of the cigarette end face; The physical defect detection results of the cigarette end face are obtained based on the geometric parameters of the cigarette end face.

2. The detection method according to claim 1, characterized in that, The reference plane for the cigarette end face is obtained based on the three-dimensional point cloud data, including: The three-dimensional point cloud data is preprocessed, including voxel raster simplification and statistical filtering. Based on the preprocessed 3D point cloud data, RANSAC was used to extract the reference plane of the cigarette end face.

3. The detection method according to claim 2, characterized in that, Based on the preprocessed 3D point cloud data, the reference plane of the cigarette end face was extracted using RANSAC, including: Three non-collinear points are randomly selected from the preprocessed 3D point cloud data; A planar model is built based on three non-collinear points extracted; Calculate the distance from each point in the preprocessed 3D point cloud data to the planar model; Each point is classified according to the calculated distance to obtain the set of interior points and the number of interior points in the set of interior points; Determine whether the iteration is complete; If the iteration is not completed, return to the step of randomly selecting three non-collinear points from the preprocessed 3D point cloud data; If the iteration is completed, the plane model corresponding to the maximum number of internal points is selected as the reference plane for the cigarette end face.

4. The detection method according to claim 1, characterized in that, The calculation of the geometric parameters of the cigarette end face based on the reference plane of the cigarette end face includes: Extract point cloud data of tobacco shreds inside the end face of the cigarette stick; The cavitation depth of the tobacco shreds is calculated according to formula (1). ，（1） in, The depth of the cavitation. The height of the reference plane at the end face of the cigarette stick. For the first The depth of the cavity filled with tobacco inside the block This represents the number of void defects.

5. The detection method according to claim 1, characterized in that, The calculation of the geometric parameters of the cigarette end face based on the reference plane of the cigarette end face includes: Fit the two-dimensional projection of the outer ring of the cigarette wrapping paper to obtain the total area of ​​the cigarette end face; Extract the three-dimensional point cloud data at a preset depth below the reference plane of the cigarette end face, and project it onto the XOY plane to obtain a two-dimensional image; The cavitation section ratio is calculated according to formula (2). ，（2） in, The ratio of the void section, The second in the two-dimensional image The area of ​​the vacant region, This represents the number of vacant areas. The total area of ​​the cigarette end face.

6. The detection method according to claim 1, characterized in that, The calculation of the geometric parameters of the cigarette end face based on the reference plane of the cigarette end face includes: The point cloud of the cigarette end face is sliced ​​along the Z-axis at a preset height step to obtain multiple slice planes parallel to the reference plane; The point cloud data of each slice plane is projected onto the corresponding two-dimensional raster image to obtain the void region; For each of the aforementioned void regions, extract its convex hull profile on the XOY plane and calculate the cross-sectional area enclosed by the convex hull profile. Calculate the void volume according to formula (3). ，（3） in, For the volume of the void, This represents the number of slice layers. For the first The height of the slice, For the first The cross-sectional area enclosed by the convex hull profile of the layer. This is the preset height step size.

7. The detection method according to claim 1, characterized in that, The calculation of the geometric parameters of the cigarette end face based on the reference plane of the cigarette end face includes: Based on the three-dimensional point cloud data, the outermost circumferential boundary of the cigarette end face is obtained by fitting. Detection is performed along the circumferential boundary direction to identify continuous point cloud segments that are lower than the reference plane by more than a preset threshold in the height direction, and these segments are determined to be contact areas; In the point cloud data of each contact area, the point with the smallest Z coordinate is selected as the lowest contact point; Calculate the contact depth according to formula (4). ，（4） in, For contact depth, , , , For the plane parameters of the reference plane, This is the lowest contact point; The three-dimensional point cloud data is projected onto the XOY plane to obtain a two-dimensional image of the cigarette end face. Extract the contact area from the two-dimensional image of the cigarette end face; Calculate the ratio of the total contact length to the circumference using formula (5). ，（5） in, The ratio of the total length of the contact to the circumference. For the first The central angle radians corresponding to each contact. This refers to the number of contacts.

8. The detection method according to claim 1, characterized in that, The calculation of the geometric parameters of the cigarette end face based on the reference plane of the cigarette end face includes: Extract point cloud data higher than the reference plane of the cigarette end face from the point cloud data of the cigarette end face, and use it as the point cloud data for cigarette spinning; Obtain the maximum value of the point cloud data in the Z-axis direction; Calculate the silk-spinning length according to formula (6). ，（6） in, The length of the silk thread. This represents the maximum value of the point cloud data along the Z-axis. The height of the reference plane at the end face of the cigarette stick; Calculate the average height of the silk-spinning point cloud data along the Z-axis; Calculate the end face flatness according to formula (7). ，（7） in, For end face flatness, The first point cloud data for spinning silk The height value of each point. The average height of the point cloud data for spinning silk. This represents the total number of points in the point cloud data.

9. An automatic detection system for physical defects on the end face of a cigarette stick, characterized in that, The system includes: A line structured light 3D imaging module is used to acquire 3D point cloud data of the end face of the cigarette stick to be tested; A precision moving platform is used to support and fix the cigarette to be inspected, so that the end face of the cigarette is kept within the appropriate measurement range of the line structured light 3D imaging device; The control module is used to control the precision moving platform to move along a set trajectory; The data processing module is used to obtain the reference plane of the cigarette end face based on the three-dimensional point cloud data, calculate the geometric parameters of the cigarette end face, and output the detection results of physical defects of the cigarette end face. A processor configured to perform the method as described in any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions that, when executed by a processor, implement the method as described in any one of claims 1 to 8.

Images