Method, device, system and storage medium for detecting coil tower defects of steel coils

Abstract

This invention provides a method, device, system, and storage medium for detecting tower-shaped defects in steel coils. The detection method includes: acquiring a point cloud set of the steel coil; segmenting a steel coil point cloud model from the point cloud set using a 3D point cloud algorithm; extracting planar regions from the steel coil point cloud model; calculating the vertical distance from the point cloud to the planar regions and applying a threshold to extract the target point cloud; and performing principal component analysis on the target point cloud to determine whether the steel coil has a tower-shaped defect. This method adds an extra dimension compared to current image recognition methods, providing more accurate direct 3D coordinate information. It is less affected by changes in external lighting and imaging distance. Based on the characteristics of tower-shaped defects, precise 3D information is required for this defect. By judging the magnitude of the overflow distance, it is determined whether the steel coil needs rework. This method enables fully automated identification of steel coil defects, reducing manpower and material resources and improving product yield.

Description

Technical Field

[0001] This invention belongs to the field of automated processing, and in particular relates to a method, equipment, system and storage medium for detecting defects in steel coil towers. Background Technology

[0002] In traditional steel mills, during the coiling process, manufacturing issues may cause uneven ends of cold-rolled strip steel sheets (coils), resulting in some coils being higher than others, exhibiting varying degrees of pagoda or bread-like shapes. This condition is known as a steel coil pagoda-shaped defect.

[0003] Defects in the coil shape can significantly impact subsequent processing stages, drastically reducing the quality of steel coil products. Currently, traditional manufacturers rely on manual inspection. However, because coil shape defects are often subtle, manual identification is prone to errors, making it impossible to promptly address specific defects. Some manufacturers use image recognition or 3D point cloud methods to determine the coil shape. However, image recognition, relying on cameras, cannot capture accurate 3D information. Determining whether rework is necessary by judging the height of the coil protrusion from the coil's end face is often inaccurate. Point cloud acquisition requires tightly designed fixtures, such as mounting brackets to secure the coil during unwinding and precise positioning, which is costly and geographically limited.

[0004] Therefore, there is an urgent need for a new method for detecting defects in steel coil towers. Summary of the Invention

[0005] To address the technical problems identified in the prior art, this invention provides a method, equipment, system, and storage medium for detecting defects in steel coil towers.

[0006] The first aspect of this application provides a method for detecting tower-shaped defects in steel coils. The method includes: acquiring a point cloud set of steel coils; segmenting a steel coil point cloud model from the point cloud set using a 3D point cloud algorithm; extracting planar regions from the steel coil point cloud model; calculating the vertical distance from the point cloud to the planar region in the steel coil point cloud model and performing a threshold judgment to extract a target point cloud; and performing principal component analysis on the target point cloud to determine whether the steel coil has a tower-shaped defect.

[0007] In a further embodiment of this application, segmenting a steel coil point cloud set into a steel coil point cloud model using a 3D point cloud algorithm includes: segmenting the steel coil point cloud model from the steel coil point cloud set by fitting a cylinder; or using a clustering segmentation algorithm based on RGB information to obtain the steel coil point cloud model from the steel coil point cloud set.

[0008] In a further embodiment of this application, calculating the vertical distance from the point cloud in the steel coil point cloud model to the planar region and performing a threshold judgment to extract the target point cloud includes: segmenting the cylindrical surface point cloud of the steel coil according to the diameter information and center point coordinates of the steel coil and deleting the cylindrical surface point cloud to leave the remaining point cloud to be calculated; obtaining the vertical distance from the point cloud to be calculated to the planar region through a distance algorithm; and extracting the target point cloud whose vertical distance is greater than a preset distance threshold.

[0009] In a further embodiment of this application, obtaining the vertical distance from the point cloud to the planar region using a distance algorithm includes: dividing the planar region into a predetermined first number of sectors along the circumference; and calculating the vertical distance from the point cloud to the planar region within each sector.

[0010] In a further aspect of this application, performing principal component analysis on the target point cloud to determine whether the steel coil has a tower-shaped defect includes: performing principal component analysis on the target point cloud to obtain the side length enclosed by the target point cloud; and determining that the steel coil has a tower-shaped defect when the side length exceeds a preset side length threshold.

[0011] In a further embodiment of this application, the detection method further includes, before acquiring the point cloud of the steel coil: acquiring a frame of end face image of the steel coil; calculating the roundness of the end face image; adjusting the direction of the point cloud acquisition device according to the roundness, cyclically acquiring the end face image and calculating the roundness until the roundness meets the preset roundness threshold; calculating the center point coordinates of the steel coil; and moving the point cloud acquisition device to the center point coordinates to complete the initialization.

[0012] In a second aspect, this application also provides a detection device for tower-shaped defects in steel coils. The detection device includes: a point cloud acquisition device for acquiring point clouds of steel coils; and a host computer electrically connected to the point cloud acquisition device, configured to: control the point cloud acquisition device to acquire the point clouds of steel coils; segment the steel coil point cloud model from the point clouds of steel coils using a 3D point cloud algorithm; extract planar regions from the steel coil point cloud model; calculate the vertical distance from the point clouds in the steel coil point cloud model to the planar regions and perform threshold judgment to extract target point clouds; and perform principal component analysis on the target point clouds to determine whether tower-shaped defects exist in the steel coils.

[0013] In a further embodiment of this application, the detection device further includes: a PWM controller for controlling the drive device via pulse width modulation; a drive device connected to one side of the point cloud acquisition device for driving the point cloud acquisition device to rotate / move; an image acquisition device located on one side of the point cloud acquisition device for acquiring images; and the host computer is further configured to: control the image acquisition device to acquire a frame of end face image of the steel coil before acquiring the point cloud set; calculate the roundness of the end face image; control the drive device to adjust the direction of the point cloud acquisition device, cyclically acquire end face images and calculate roundness until the roundness meets a preset roundness threshold; calculate the center point coordinates of the steel coil; and move the point cloud acquisition device to the center point coordinates to complete initialization.

[0014] A third aspect of this application also provides a detection system for tower-shaped defects in steel coils, comprising: a point cloud acquisition module for receiving and acquiring a point cloud set of steel coils and segmenting a steel coil point cloud model from the point cloud set using a 3D point cloud algorithm; an extraction module for extracting planar regions from the steel coil point cloud model; a calculation module for calculating the vertical distance from the point cloud to the planar region in the steel coil point cloud model and performing a threshold judgment to extract a target point cloud; a judgment module for performing principal component analysis on the target point cloud to determine whether the steel coil has a tower-shaped defect; and an initialization module for receiving an end face image of the steel coil; calculating the roundness of the end face image; controlling a drive device to adjust the direction of the point cloud acquisition device, cyclically acquiring end face images and calculating roundness until the roundness meets a preset roundness threshold, and then stopping the control of the point cloud acquisition device to move; calculating the center point coordinates of the steel coil; and moving the point cloud acquisition device to the center point coordinates to complete the initialization.

[0015] Finally, this application also provides a computer-readable storage medium storing one or more programs, which can be executed by one or more processors to implement the detection method described above. Beneficial effects

[0016] This invention provides a method for detecting tower-shaped defects in steel coils. This method involves establishing a point cloud model of the steel coil, segmenting its planar regions (corresponding to the two end faces of the coil) within the model, calculating the vertical distance from the point cloud to the planar regions, and applying a threshold to extract the target point cloud. Principal component analysis is then performed on the target point cloud to determine if a tower-shaped defect exists in the steel coil. This method adds an extra dimension compared to current image recognition methods, providing more accurate direct 3D coordinate information. It is less affected by changes in external lighting and imaging distance. Based on the characteristics of tower-shaped defects, precise 3D information is required. The magnitude of the overflow distance is used to determine whether rework of the steel coil is necessary. This method enables fully automated identification of steel coil defects, reducing manpower and material resources and improving product yield.

[0017] Furthermore, this invention improves the reliability of identifying steel coil tower-shaped defects by configuring the point cloud acquisition device with rotational and translational degrees of freedom and integrating 2D image recognition. This ensures the accuracy of the origin coordinate system established by the point cloud acquisition device. Simultaneously, the steel coil is not limited by environment or site, eliminating the need for sophisticated tooling fixtures and offering greater versatility.

[0018] Other features and advantages of the embodiments of the present invention will be described in the following detailed description section. Attached Figure Description

[0019] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0020] Figure 1 A flowchart illustrating the method for detecting defects in steel coil towers provided in an embodiment of the present invention;

[0021] Figure 2 This is a flowchart illustrating step S4 in the method for detecting defects in steel coil towers provided in this embodiment of the invention.

[0022] Figure 3 This is a flowchart illustrating step S42 of the method for detecting defects in steel coil towers provided in this embodiment of the invention.

[0023] Figure 4 This is a flowchart illustrating step S5 in the method for detecting defects in steel coil towers provided in this embodiment of the invention.

[0024] Figure 5 This is another flowchart of the method for detecting defects in steel coil towers provided in an embodiment of the present invention;

[0025] Figure 6 A logic diagram of the method for detecting defects in steel coil towers provided in this embodiment of the invention; and

[0026] Figure 7 This is a connection topology diagram of the detection device provided in an embodiment of the present invention;

[0027] Figure 8 This is a schematic diagram of the detection system provided in an embodiment of the present invention.

[0028] Attached Figure

[0029] 100. Detection equipment; 101. Point cloud acquisition device;

[0030] 102. Host computer; 103. PWM controller;

[0031] 104. Driving device; 105. Image acquisition device;

[0032] 200. Detection system; 201. Point cloud acquisition module;

[0033] 202. Extraction module; 203. Calculation module;

[0034] 204. Judgment module; 205. Initialization module. Detailed Implementation

[0035] To make the above and other features and advantages of the present invention clearer, the invention will be further described below with reference to the accompanying drawings. It should be understood that the specific embodiments given herein are for the purpose of explanation to those skilled in the art and are exemplary only, not restrictive.

[0036] In order to address the technical problems existing in the prior art, whether using manual methods, image recognition, or 3D depth recognition, there are technical defects in the accuracy of the methods. The embodiments of the present invention provide a general inventive concept, namely, a method for detecting defects in steel coil towers.

[0037]

Detection Method

[0038] Please see Figures 1-2 , Figure 1 A flowchart illustrating the method for detecting defects in steel coil towers provided in an embodiment of the present invention;

[0039] This invention provides a method for detecting defects in steel coil towers, the method comprising the following steps:

[0040] Step S1: Obtain the point cloud of the steel coil;

[0041] Step S2: Segment the steel coil point cloud model from the steel coil point cloud set using a 3D point cloud algorithm;

[0042] Step S3: Extract the planar regions from the steel coil point cloud model;

[0043] Step S4: Calculate the vertical distance from the point cloud to the planar region in the steel coil point cloud model and perform threshold judgment to extract the target point cloud;

[0044] Step S5: Perform principal component analysis on the target point cloud to determine whether the steel coil has a tower-shaped defect.

[0045] It is understood that steps S1 to S2 describe obtaining a steel coil point cloud model from point cloud data. In this embodiment of the invention, the point cloud is a set of points. Compared to an image point cloud, it includes depth parameters. In other words, the 3D point cloud directly provides 3D spatial data. Compared to image recognition in the prior art, which requires perspective geometry to infer 3D data, the data derived from perspective geometry may not meet the requirements and may contain some deviations. Therefore, using a steel coil point cloud model can significantly improve the accuracy of detection.

[0046] In one specific scheme, a laser radar is used to scan the steel coils along a preset route to obtain a point cloud set of steel coils; point cloud features are extracted from the point cloud set of steel coils based on the characteristics of the steel coils; and a clustering segmentation algorithm is used to obtain a point cloud model of the steel coils.

[0047] It is understandable that the steel coil point cloud model in step S1 needs to extract targets from the collected steel coil point cloud set. The two key steps of target extraction are: feature extraction and selection, and classification. By scanning the steel coils sequentially according to the scanning route using a LiDAR, the overall steel coil point cloud set can be obtained. Since the steel coils to be identified are usually in an environment where various point clouds are mixed together, when identifying targets in the steel coil point cloud model, there needs to be an index or value to maximize the difference between the steel coil point cloud model and other point clouds, i.e., the target features mentioned above.

[0048] In this application, cylindrical fitting is preferably used for segmentation. The SACMODEL_CYLINDER model (a cylinder defined in the point cloud library) provided by PCL (Point Cloud Library) is used for cylindrical segmentation via a stochastic parameter estimation method, thereby extracting the steel coil point cloud model. For example, the steel coil point cloud set is loaded, and the normal is iteratively calculated using the stochastic parameter estimation method to perform cylindrical segmentation.

[0049] In another feasible example, each point cloud may include RGB information in addition to three-dimensional coordinates (x, y, z). This information can be extracted based on the point cloud color. First, the steel coil point cloud set is preprocessed by filtering and downsampling. Then, clustering and segmentation are performed based on the color of the steel coil to obtain the steel coil point cloud model.

[0050] It is understandable that the method of target extraction is not restricted. Since the steel coil structure is a simple column, other methods such as edge extraction can be used to mainly remove background noise in the steel coil point cloud to obtain the steel coil point cloud model.

[0051] In step S3, planar regions are extracted from the steel coil point cloud model. Specifically, this is done by using tangent vectors on the same plane that share the same characteristics, such as the slope of the tangent line y'=frac{y'}{1} of the planar curve y=g(x), where the tangent vectors are all (1,y'). Clustering can be performed based on this characteristic. Specifically, two tangent vectors are created in the x and y directions of the steel coil point cloud model (the axial direction of the steel coil is the y direction, and the direction perpendicular to the axial direction is the x direction). The tangent vectors are calculated using an integral image. The point cloud normals in the steel coil point cloud model are calculated based on the two tangent vectors. The points in the point cloud normals are projected onto a spherical coordinate system for re-clustering to obtain candidate plane clusters. Finally, planes with similar local surface normals are clustered in the distance space (distance between the plane and the origin), thereby extracting the planar regions in the steel coil point cloud model and establishing the plane equations corresponding to these planar regions.

[0052] In another optional scheme in step S3, the feature that the distance from a point to a plane is equal can also be used for plane fitting. A point is randomly selected from the end face of the steel coil in the steel coil point cloud model, along with the nearest point to that point, to form a point set. The target point closest to this point set is automatically found, and the distance and mean square error from the target point to the preset coordinate system YZ plane are calculated. If the distances are equal, the point is added to the current point set. If the number of points in the point set is greater than a threshold, it is considered valid (proving that a plane exists). This process continues until all points have been traversed to obtain all point clouds of the same plane, and the plane equation is established.

[0053] Please see Figure 2 , Figure 2 This is a flowchart illustrating step S4 of the steel coil tower-shaped defect detection method provided in this embodiment of the invention. Step S4 involves calculating the vertical distance from the point cloud in the steel coil point cloud model to the planar region and performing a threshold judgment to extract the target point cloud, including:

[0054] Step S41: Based on the diameter information and center point coordinates of the steel coil, segment the cylindrical surface point cloud of the steel coil and delete the cylindrical surface point cloud to leave the remaining point cloud to be calculated.

[0055] Step S42: Obtain the vertical distance from the point cloud to the planar region using a distance algorithm;

[0056] Step S43: Extract the target point cloud with a vertical distance greater than a preset distance threshold.

[0057] It is understandable that in step S41, the point cloud to be calculated can be segmented according to the diameter information and center point coordinates of the steel coil. For example, if the inner diameter and outer diameter of the steel coil are d1 and d2 respectively, and the center point coordinate is i, the first circumference and the second circumference are drawn around i with diameters d1+△l and d1-△l respectively. The point cloud within the coordinate system of the first circumference and the second circumference are filtered out, thereby eliminating the point cloud on the inner surface of the cylinder. The same applies to the outer surface, so that only the remaining point cloud to be calculated is left in the steel coil point cloud model.

[0058] At this point, the distance from the remaining point cloud to the planar region is calculated using the perpendicular distance formula from the point to the plane equation, and all remaining point clouds to be calculated are traversed.

[0059] Finally, target point clouds with a vertical distance greater than the distance threshold are extracted. Generally speaking, the distance threshold is set at around 6cm according to current process requirements.

[0060] The purpose of step S41 is to reduce computing power and improve response speed.

[0061] Please continue reading. Figure 3 , Figure 3This is a flowchart illustrating step S42 of the steel coil tower-shaped defect detection method provided in this embodiment of the invention. Step S42, which involves obtaining the vertical distance from the point cloud to the planar region using a distance algorithm, includes:

[0062] Step S421: Divide the planar region into a predetermined first number of sectors along the circumference;

[0063] Step S422: Calculate the vertical distance from the point cloud to the planar region within each sector.

[0064] It is understandable that the purpose of steps S421 to S422 is twofold: firstly, to improve the calculation speed, and secondly, to facilitate subsequent principal component analysis.

[0065] Please see Figure 4 , Figure 4 This is a flowchart illustrating step S5 of the method for detecting tower-shaped defects in steel coils provided in this embodiment of the invention. Step S5 involves performing principal component analysis on the target point cloud to determine whether the steel coil exhibits a tower-shaped defect, including:

[0066] Step S51: Perform principal component analysis on the target point cloud to determine the side length enclosed by the target point cloud.

[0067] Step S52: When the side length exceeds the preset side length threshold, it is determined that the steel coil has a tower-shaped defect.

[0068] This invention provides an optional method for step S5, which involves fitting the target point cloud into a line and calculating its side length. When the side length exceeds a preset side length threshold, it proves that a certain loop of the steel coil is abrupt and has a tower-shaped defect.

[0069] It is understandable that the number of sectors divided in step S421 above can be used for statistical analysis. When the number of sectors with target point clouds exceeds the set threshold, it is determined that the steel coil has a tower-shaped defect.

[0070] In summary, the present invention provides a method for detecting tower-shaped defects in steel coils. This method establishes a point cloud model of the steel coil, segments its planar regions (corresponding to the two end faces of the steel coil) within the point cloud model, calculates the vertical distance from the point cloud to the planar regions, and uses a threshold judgment to extract the target point cloud. Subsequently, principal component analysis is performed on the target point cloud to determine whether the steel coil has a tower-shaped defect. Compared with current image recognition methods, this method adds an extra dimension, providing more accurate direct 3D coordinate information. It is less affected by changes in external lighting and imaging distance. Based on the characteristics of tower-shaped defects, accurate 3D information is required for this defect. The magnitude of the overflow distance is used to determine whether the steel coil needs to be reworked. This method can achieve fully automated identification of steel coil defects, reduce manpower and material resources, and improve product yield.

[0071] Please see Figure 5 , Figure 5 This is another flowchart of the method for detecting tower-shaped defects in steel coils provided in an embodiment of the present invention. Before obtaining the steel coil point cloud in step S1, the detection method further includes:

[0072] Step S101: Obtain an image of the end face of a steel coil;

[0073] Step S102: Calculate the roundness of the end face image;

[0074] Step S103: Adjust the direction of the point cloud acquisition device according to the roundness, and repeatedly acquire end face images and calculate roundness until the roundness meets the preset roundness threshold.

[0075] Step S104: Calculate the coordinates of the center point of the steel coil;

[0076] Step S105: Move the point cloud acquisition device to the center point coordinates to complete the initialization.

[0077] It is understood that the embodiments of the present invention also provide a method for controlling the initialization of a point cloud acquisition device by combining image recognition. This method is executed before acquiring the point cloud of the steel coil to ensure the accuracy of the origin coordinate system, thereby improving the reliability of the above-mentioned method of determining whether the steel coil has a tower-shaped defect through the point cloud.

[0078] Please see Figure 6 , Figure 6 This is a logic diagram of the method for detecting tower-shaped defects in steel coils provided in this embodiment of the invention. Specifically, its motion control can employ a mobile multi-axis robotic arm, with the image acquisition device and point cloud acquisition device coaxially mounted at the end of the robotic arm. Before each point cloud determination, the robotic arm is first moved to the end face of the steel coil, and then the following steps are executed:

[0079] Step A: Obtain a frame of steel coil image, then proceed to Step B;

[0080] Step B: Segment the end face image of the steel coil from the coil image;

[0081] Step C: Calculate the roundness of the end face image;

[0082] Step D: Perform a roundness threshold judgment on the roundness. If the roundness meets the roundness threshold range, proceed to step E; if the roundness does not meet the roundness threshold range, proceed to step G.

[0083] Step E: Calculate the coordinates of the center point of the end face image;

[0084] Step F: Move the point cloud acquisition device to the center point coordinates;

[0085] Step G: Adjust the angle of the point cloud acquisition device; cycle through step A.

[0086] Step A involves acquiring a frame of steel coil image, which includes a background and a target. Step B uses edge detection segmentation to separate the end face image of the steel coil from the background, thus completing feature extraction.

[0087] In step C, through image recognition, various feature data of the current steel coil end face are extracted, and the roundness can be calculated using 4*PI*A / P^2; where PI represents Π, A represents the area of ​​the region, and P represents the perimeter of the region.

[0088] Then, step D is executed to determine the roundness threshold of the acquired roundness value. When the roundness meets the roundness threshold, it is determined that the current point cloud acquisition device is aligned. At this time, the central axis of the point cloud acquisition device moves to align with the calculated center point coordinates, and the initialization is completed.

[0089] If the roundness does not meet the roundness threshold, steps A to D are repeated by controlling the angle of the laser emitter of the point cloud acquisition device.

[0090] If the measured roundness does not meet the roundness threshold, when adjusting the angle, first adjust the angle in the first direction. If the roundness changes in the direction that deviates from the roundness threshold, then adjust the angle in the opposite direction of the first direction the next time you adjust the angle, until the measured roundness meets the roundness threshold.

[0091] Furthermore, step S103 (step G) also includes:

[0092] Calculate the difference between the currently measured roundness and the roundness threshold, and adjust the angle value that the point cloud acquisition device needs to adjust based on the difference.

[0093] Specifically, in this embodiment of the invention, the angle of the point cloud acquisition device is controlled by PWM pulse control. Each time, the preset angle can be adjusted according to the difference. When the difference between the currently measured roundness and the roundness threshold is large, the point cloud acquisition device can be rotated and adjusted by 5° to 15°. When the difference is small, the angle value can be 1° to 5°. When the difference is close to 0, fine adjustment begins, and the angle can be set below 1°.

[0094] Furthermore, after the laser acquisition head of the point cloud acquisition device is aligned with the center point coordinates, the point cloud acquisition device is moved in the direction of the steel coil end face, that is, in the direction that the laser acquisition head is currently facing, until the laser acquisition head and the end face are on the same horizontal plane. At this time, the end of the laser acquisition head is used as the origin coordinate system.

[0095] It is understandable that by configuring the point cloud acquisition device to have rotational and translational degrees of freedom and integrating 2D image recognition, the origin coordinate system established by the point cloud acquisition device can be made more accurate, thereby improving the reliability of judging steel coil tower defects through 3D point clouds. At the same time, steel coils are not limited by environment and site, and do not require the design of precise tooling fixtures, thus having greater versatility.

[0096] Testing Equipment

[0097] Please see Figure 7 , Figure 7 This is a connection topology diagram of the detection device provided in an embodiment of the present invention.

[0098] A second aspect of this application also provides a detection device 100 for defects in steel coil towers, the detection device comprising:

[0099] The point cloud acquisition device 101 includes a laser acquisition head for acquiring point clouds of steel coils;

[0100] The host computer 102 is electrically connected to the point cloud acquisition device 101. The host computer 102 contains a program that is compiled and configured to execute the above-mentioned detection method for steel coil tower-shaped defects.

[0101] It is understandable that the point cloud acquisition device 101 can obtain more accurate three-dimensional information than images, which has advantages in identifying defects in steel coil towers. At the same time, the above method can reduce manpower and material resources and improve the degree of automation.

[0102] In this embodiment of the invention, the detection device 100 further includes a PWM controller 103, a drive device 104, and an image acquisition device 105;

[0103] The PWM controller 103 is used to control the drive device 104 through pulse width modulation. The drive device 104 may be a robotic arm. The point cloud acquisition device 101 is connected to the end of the drive device 104. The drive device 104 is used to drive the point cloud acquisition device 101 to rotate / move.

[0104] The image acquisition device 105, like the point cloud acquisition device 101, is located at the end of the driving device 104 and is used to acquire two-dimensional images.

[0105] Before acquiring the coil point aggregation, the host computer was also configured to:

[0106] Control the image acquisition device to acquire a frame of end face image of the steel coil;

[0107] Calculate the roundness of the end face image;

[0108] The control drive device adjusts the direction of the point cloud acquisition device, cyclically acquires end face images and calculates roundness until the roundness meets the preset roundness threshold.

[0109] Calculate the coordinates of the center point of the steel coil;

[0110] The mobile point cloud acquisition device has completed initialization by moving to the center point coordinates.

[0111] As mentioned above, by configuring the point cloud acquisition device to have rotational and translational degrees of freedom and integrating 2D image recognition, the origin coordinate system established by the point cloud acquisition device is accurate, thereby improving the reliability of judging steel coil tower defects through 3D point clouds. At the same time, steel coils are not limited by environment and site, and do not require the design of precise tooling fixtures, thus possessing greater versatility.

[0112] [Detection System]

[0113] Please see Figure 8 , Figure 8 This is a schematic diagram of the modules of the detection system 200 provided in an embodiment of the present invention.

[0114] This invention also provides a detection system 200 for defects in steel coil towers, comprising:

[0115] Point cloud acquisition module 201 is used to receive and acquire the point cloud set of steel coils and segment the steel coil point cloud model from the point cloud set using a 3D point cloud algorithm.

[0116] Extraction module 202 is used to extract planar regions from the point cloud model of the steel coil;

[0117] Calculation module 203 is used to calculate the vertical distance from the point cloud to the planar region in the steel coil point cloud model and to perform threshold judgment to extract the target point cloud;

[0118] The judgment module 204 is used to perform principal component analysis on the target point cloud to determine whether the steel coil has a tower-shaped defect;

[0119] Initialization module 205 is used to receive the end face image of the steel coil; calculate the roundness of the end face image; control the drive device to adjust the direction of the point cloud acquisition device, cyclically acquire the end face image and calculate the roundness until the roundness meets the preset roundness threshold, then stop controlling the movement of the point cloud acquisition device; calculate the center point coordinates of the steel coil; and move the point cloud acquisition device to the center point coordinates to complete the initialization.

[0120] It is understandable that the above detection methods are encapsulated in different modules through programming to build a system. The beneficial effects are the same as described above, and will not be repeated here.

[0121] Furthermore, those skilled in the art should understand that if all or part of the sub-modules involved in the detection system 200 provided in the embodiments of the present invention are combined or replaced by means of fusion, simple changes, mutual transformation, etc., such as moving the position of each component; or setting the product they constitute as a whole; or having a detachable design; as long as the combined components can form a device / apparatus / system with a specific function, using such a device / apparatus / system to replace the corresponding components of the present invention also falls within the protection scope of the present invention.

[0122] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0123] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A method for detecting a coil tower type defect of a steel coil, characterized by, The detection method includes: Steel coil acquisition points are gathered; The point cloud model of the steel coil was segmented from the point cloud set of the steel coil using a 3D point cloud algorithm; Extract the planar regions from the point cloud model of the steel coil; Calculate the vertical distance from the point cloud to the planar region in the steel coil point cloud model and perform threshold judgment to extract the target point cloud; Principal component analysis was performed on the target point cloud to determine whether the steel coil had a tower-shaped defect. The process of calculating the vertical distance from the point cloud to the planar region in the steel coil point cloud model and performing a threshold judgment to extract the target point cloud includes: segmenting the cylindrical surface point cloud of the steel coil according to the diameter information and center point coordinates of the steel coil and deleting the cylindrical surface point cloud to leave the point cloud to be calculated; obtaining the vertical distance from the point cloud to be calculated to the planar region through a distance algorithm; and extracting the target point cloud whose vertical distance is greater than a preset distance threshold. The step of performing principal component analysis on the target point cloud to determine whether the steel coil has a tower-shaped defect includes: performing principal component analysis on the target point cloud to obtain the side length enclosed by the target point cloud; when the side length exceeds a preset side length threshold, it is determined that the steel coil has a tower-shaped defect. The detection method further includes, before acquiring the point cloud of the steel coil: acquiring a frame of end face image of the steel coil; calculating the roundness of the end face image; adjusting the direction of the point cloud acquisition device according to the roundness, cyclically acquiring the end face image and calculating the roundness until the roundness meets the preset roundness threshold; calculating the center point coordinates of the steel coil; and moving the point cloud acquisition device to the center point coordinates to complete the initialization.

2. The detection method according to claim 1, characterized in that, The step of segmenting the steel coil point cloud model from the steel coil point cloud set using a 3D point cloud algorithm includes: A steel coil point cloud model is segmented from the steel coil point cloud set using cylindrical fitting; or Based on RGB information, a clustering segmentation algorithm is used to obtain a steel coil point cloud model from the point cloud dataset.

3. The method of claim 1, wherein The step of obtaining the vertical distance from the point cloud to the planar region using a distance algorithm includes: The planar region is divided into a predetermined first number of sectors along the circumference; Calculate the vertical distance from the point cloud to be calculated within each sector to the planar region.

4. An apparatus for detecting a coil tower type defect of a steel coil, characterized by, The detection equipment includes: Point cloud acquisition device, used to collect point clouds of steel coils; The host computer is electrically connected to the point cloud acquisition device, and the host computer is configured to: The control point cloud acquisition device acquires the point cloud set of the steel coil; The point cloud model of the steel coil was segmented from the point cloud set of the steel coil using a 3D point cloud algorithm; Extract the planar regions from the point cloud model of the steel coil; Calculate the vertical distance from the point cloud to the planar region in the steel coil point cloud model and perform threshold judgment to extract the target point cloud; Principal component analysis was performed on the target point cloud to determine whether the steel coil had a tower-shaped defect. The process of calculating the vertical distance from the point cloud to the planar region in the steel coil point cloud model and performing a threshold judgment to extract the target point cloud includes: segmenting the cylindrical surface point cloud of the steel coil according to the diameter information and center point coordinates of the steel coil and deleting the cylindrical surface point cloud to leave the point cloud to be calculated; obtaining the vertical distance from the point cloud to be calculated to the planar region through a distance algorithm; and extracting the target point cloud whose vertical distance is greater than a preset distance threshold. The step of performing principal component analysis on the target point cloud to determine whether the steel coil has a tower-shaped defect includes: performing principal component analysis on the target point cloud to obtain the side length enclosed by the target point cloud; when the side length exceeds a preset side length threshold, it is determined that the steel coil has a tower-shaped defect. The detection equipment also includes: A PWM controller is used to control a drive device via pulse width modulation. A drive unit, connected to one side of the point cloud acquisition device, is used to drive the point cloud acquisition device to rotate / move. An image acquisition device is disposed on one side of the point cloud acquisition device and is used to acquire images; The host computer is also configured to: Control the image acquisition device to acquire a frame of end face image of the steel coil; Calculate the roundness of the end face image; The control drive device adjusts the direction of the point cloud acquisition device, cyclically acquires end face images and calculates roundness until the roundness meets the preset roundness threshold. Calculate the coordinates of the center point of the steel coil; Move the point cloud acquisition device to the center point coordinates to complete initialization.

5. A system for detecting a coil tower type defect, characterized by, include: The point cloud acquisition module is used to receive and acquire the point cloud set of steel coils and segment the steel coil point cloud model from the point cloud set using a 3D point cloud algorithm. The extraction module is used to extract planar regions from the point cloud model of the steel coil; The calculation module is used to calculate the vertical distance from the point cloud to the planar region in the steel coil point cloud model and to perform threshold judgment to extract the target point cloud; The judgment module is used to perform principal component analysis on the target point cloud to determine whether the steel coil has a tower-shaped defect. An initialization module is used to receive images of the end face of the steel coil; Calculate the roundness of the end face image; control the driving device to adjust the direction of the point cloud acquisition device, cyclically acquire the end face image and calculate the roundness until the roundness meets the preset roundness threshold, then stop controlling the movement of the point cloud acquisition device; calculate the center point coordinates of the steel coil; move the point cloud acquisition device to the center point coordinates to complete the initialization; The process of calculating the vertical distance from the point cloud to the planar region in the steel coil point cloud model and performing a threshold judgment to extract the target point cloud includes: segmenting the cylindrical surface point cloud of the steel coil according to the diameter information and the center point coordinates, and deleting the cylindrical surface point cloud to leave the remaining point cloud to be calculated; obtaining the vertical distance from the point cloud to be calculated to the planar region through a distance algorithm; and extracting the target point cloud whose vertical distance is greater than a preset distance threshold. The step of performing principal component analysis on the target point cloud to determine whether the steel coil has a tower-shaped defect includes: performing principal component analysis on the target point cloud to obtain the side length enclosed by the target point cloud; when the side length exceeds a preset side length threshold, it is determined that the steel coil has a tower-shaped defect.

6. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores one or more programs, and one or more of the programs can be executed by one or more processors to implement the detection method as described in any one of claims 1 to 3.

Images