A method for detecting the rotating state of a slab during the process of turning the slab based on machine vision

Abstract

The application discloses a slab rotation state detection method based on machine vision in a process of turning a steel, which comprises the following steps: acquiring multiple frame image information of a slab on a turning roller bed by using a camera device; acquiring contour information of the slab through the image information; recording position information of each frame slab based on preset mark points on the contour information; according to a sequence of the multiple frames, comparing positions between adjacent two frames, outputting a rotation direction and a rotation angle corresponding to the slab, and assisting in determining parameter details describing a rotation state of the slab, so that the rotation direction and the rotation angle of the slab are obtained. Thus, the effect of automatically monitoring the rotation state of the slab in the process of turning the steel is achieved.

Description

Technical Field

[0001] This invention relates to the technical field of automatic steel transfer, and more particularly to a method for detecting the rotation state of slabs during the steel transfer process based on machine vision. Background Technology

[0002] Slab rotation, or slab turning, involves the slab resting on a rotating roller conveyor. The purpose of this rotation is to control the width of the finished product. Based on the required width specifications, a model is used to set a specific pass for rotation, rotating the steel plate 90° to interchange its length and width, thus controlling the final width of the finished steel plate. Currently, the slab rotation process is still manually controlled, directly impacting the automation of the entire production line in the rolling of medium and heavy plates. The main challenge in automating slab rotation lies in monitoring the process and determining the timing of its termination. Therefore, a stable, accurate, and adaptable method for detecting the slab rotation status is needed. Summary of the Invention

[0003] This application provides a method for detecting the rotation state of a slab during steel transfer based on machine vision, which at least partially solves the technical problem of manual control of steel transfer in the prior art, and achieves the technical effect of automatically monitoring the rotation state of the slab during steel transfer.

[0004] Firstly, to solve the above-mentioned technical problems, embodiments of the present invention provide the following technical solutions:

[0005] A machine vision-based method for detecting the rotation state of a slab during steel transfer includes:

[0006] Multiple frames of images of slabs on the steel transfer roller conveyor are acquired using camera equipment;

[0007] The outline information of the slab is obtained through the above image information;

[0008] Based on the above contour information, preset marker points are used to record the position information of each frame of slab;

[0009] Based on the above sequence of multiple frames, the rotation direction and rotation angle corresponding to the slab are output by comparing the positions between two adjacent frames.

[0010] Optionally, the step of obtaining the contour information of the slab through the image information further includes:

[0011] Semantic segmentation techniques were used to extract the slab region and background region from the above image information;

[0012] The slab region is denoised, and a minimum bounding rectangle is established based on the contour of the slab region.

[0013] Optionally, after establishing the minimum bounding rectangle based on the contour within the slab region, the method may include:

[0014] When the area of ​​the slab region mentioned above accounts for less than a preset threshold in the minimum bounding rectangle;

[0015] Extract multiple straight line segments longer than a preset length from the outline of the aforementioned slab area;

[0016] Select two of the above line segments from multiple line segments and arrange them in combination. Record the two target line segments whose included angle reaches the preset angle range in each arrangement.

[0017] When the preset angle meets the first specific angle range, two corresponding straight line segments are defined to be perpendicular to each other. The two target straight line segments are extended on the slab area image, and the intersection of the extensions is taken as the vertex of the actual slab contour. When the preset angle meets the second specific angle range, two corresponding straight line segments are defined to be parallel to each other or collinear. The two straight line segments are taken as the candidate contour.

[0018] Based on the above vertex positions, the above target line segment group, the extension length, and the above candidate contours, the minimum bounding rectangle is repaired based on the preset features of the rectangle.

[0019] Optionally, after selecting two line segments as the candidate contours, the above method further includes:

[0020] Arrange and combine multiple of the above line segment groups, filter out the combination that meets the preset characteristics of a rectangle and contains the most of the above line segment groups, and determine the final circumscribed rectangle based on the above combination.

[0021] Optionally, the step of recording the position information of each frame of slab based on preset marker points on the aforementioned contour information further includes:

[0022] Using the minimum bounding rectangle as the outline, obtain the outline center of the outline;

[0023] Establish a coordinate system with the center of the outline as the origin, and obtain the coordinates of the four vertices of the smallest bounding rectangle.

[0024] Optionally, the step of outputting the rotation direction and rotation angle corresponding to the slab by comparing the positions between two adjacent frames further includes:

[0025] Based on the above sequence of multiple frames, record the initial position of the minimum bounding rectangle in the first frame;

[0026] Compare the vertex coordinates of the minimum bounding rectangle in each frame with the vertex coordinates of the previous frame, calculate the rotation angle, and determine the rotation direction based on the coordinate system.

[0027] Optionally, the step of outputting the rotation direction and rotation angle corresponding to the slab by comparing the positions between two adjacent frames further includes:

[0028] The coordinate system is established with the forward transport direction of the production line as the reference axis, based on the axis of symmetry and the center of the minimum bounding rectangle; the angle formed by the rotation of the axis of symmetry of the minimum bounding rectangle based on the reference axis is defined as the rotation angle.

[0029] If the angle is formed in the counterclockwise direction of the reference axis, the angle sign is positive; otherwise, the sign is negative.

[0030] Optionally, the step of extracting multiple straight line segments longer than a preset length on the outline of the slab region includes:

[0031] When there is a continuous straight line segment containing multiple slope differences within a second preset range, draw a straight line segment along the start and end points of the continuous straight line segment and extract it based on the preset length.

[0032] When a curve contains peak and valley values ​​within a third preset range, the curve is fitted into a straight line of the corresponding length using the least squares method, and the curve is extracted based on the preset length.

[0033] Optionally, the step of extracting multiple straight line segments longer than a preset length on the outline of the slab region further includes:

[0034] When the number of straight line segments extracted from the outline of the slab area is less than two, the extraction is deemed to have failed, the current frame is abandoned, and the process jumps to the next frame image to re-detect.

[0035] If multiple frames fail to be extracted, an alarm is issued indicating an anomaly in the current rotation state detection process.

[0036] Secondly, a machine vision-based system for detecting the rotational state of slabs during steel transfer is provided, the system comprising:

[0037] The video acquisition module is used to acquire multi-frame image information of slabs on the steel transfer roller conveyor using camera equipment;

[0038] The contour extraction module is used to obtain the contour information of the slab through the above image information;

[0039] The position acquisition module is used to record the position information of each frame of slab based on preset marker points on the above contour information;

[0040] The running status calculation module is used to output the rotation direction and rotation angle of the slab by comparing the positions between two adjacent frames based on the sequence of the above multi-frames.

[0041] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages:

[0042] First, the slab region is extracted, then the slab outline is obtained. The minimum bounding rectangle outline of the slab is estimated, and the bounding rectangle outline of the missing slab region is repaired. Rotation state analysis is then performed based on the repaired minimum bounding rectangle. The slab state is tracked based on a video frame sequence. By utilizing the changes in slab features between adjacent frames, the detailed parameters describing the slab's rotation state are determined, thereby obtaining the slab's rotation direction and angle. This achieves the effect of automatically monitoring the slab's rotation state during the steel transfer process. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1 A flowchart of a method for detecting the rotation state of a slab during steel transfer based on machine vision is provided in this application;

[0045] Figure 2 This is a schematic diagram illustrating the acquisition of slab images from the steel transfer roller conveyor in this application;

[0046] Figure 3 This is a schematic diagram illustrating the establishment of the minimum bounding rectangle based on the slab region outline in this application;

[0047] Figure 4 This is a schematic diagram of the coordinate system established based on the minimum bounding rectangle in this application;

[0048] Figure 5 This is a schematic diagram showing defects in the slab in this application;

[0049] Figure 6 This is a schematic diagram illustrating the extraction of straight line segments from the outline of the slab region in this application;

[0050] Figure 7 This is a schematic diagram of the arrangement of two sides with an included angle close to a right angle in this application;

[0051] Figure 8 This is a schematic diagram of the arrangement of two nearly collinear sides in this application;

[0052] Figure 9 This is a schematic diagram illustrating the use of translated line segments to repair the minimum bounding rectangle in this application;

[0053] Figure 10This is a schematic diagram of the final result after repair in this application;

[0054] Figure 11 This application provides a schematic diagram of a slab rotation state detection system based on machine vision during steel transfer. Detailed Implementation

[0055] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0056] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0057] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0058] In the description of this invention, it should also be noted that, unless otherwise explicitly specified and limited, the term "setup" should be interpreted broadly. For example, it can refer to a fixed setup, a detachable setup, or an integral setup; it can refer to a mechanical setup or an electrical setup; it can refer to a direct connection or an indirect connection through an intermediate medium; it can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0059] It should be understood that the embodiments of the present invention and the specific features in the embodiments are detailed descriptions of the technical solutions of this application, rather than limitations on the technical solutions of this application. Unless otherwise specified, the embodiments of the present application and the technical features in the embodiments can be combined with each other.

[0060] This application provides a machine vision-based method for detecting the rotation state of slabs during steel transfer, which improves upon the technical problem of manual control of steel transfer in the prior art and achieves the technical effect of automatically monitoring the rotation state of slabs during steel transfer.

[0061] The technical solution of this application embodiment is to solve the above-mentioned technical problems, and the general idea is as follows:

[0062] First, the slab region is extracted, then the slab outline is obtained. The minimum bounding rectangle outline of the slab is estimated, and the bounding rectangle outline of the missing slab region is repaired. Rotation state analysis is then performed based on the repaired minimum bounding rectangle. The slab state is tracked based on a video frame sequence. By utilizing the changes in slab features between adjacent frames, the detailed parameters describing the slab's rotation state are determined, thereby obtaining the slab's rotation direction and angle. This achieves the effect of automatically monitoring the slab's rotation state during the steel transfer process.

[0063] In the embodiments of this application, the following are provided: Figure 1 and Figure 2 The method shown is a machine vision-based method for detecting the rotation state of a slab during steel transfer. The method includes steps S101 to S104:

[0064] Step S101: Use a camera device to acquire multi-frame image information of the slab on the steel transfer roller conveyor;

[0065] It should be noted that, based on the steel-transfer process, a camera device is used to capture images of the slab from the top downwards to obtain images of the slab's state.

[0066] Step S102: Obtain the outline information of the slab through the above image information;

[0067] It should be noted that because the surface of the slab is flat, it is difficult to identify feature points on it. Therefore, using contour acquisition is a more effective way to monitor the state of the slab in real time.

[0068] Step S103: Record the position information of each frame of slab based on the preset marker points on the above contour information;

[0069] It should be noted that the outline of the slab is close to a rectangle. Therefore, a minimum bounding rectangle is set, and marker points (such as the vertices and center of the rectangle) are selected on the minimum bounding rectangle to record the position information of each frame of the slab.

[0070] Step S104: Based on the sequence of the above multi-frames, by comparing the positions between two adjacent frames, output the rotation direction and rotation angle corresponding to the above slab.

[0071] It should be noted that the purpose of comparing two adjacent frames is to obtain the rotation direction and rotation state of the slab in real time, so that the staff can grasp the operation of the slab throughout the entire inspection process, which provides convenience for subsequent analysis.

[0072] Furthermore, such as Figure 2 As shown, the step of obtaining the contour information of the slab through the image information further includes:

[0073] Semantic segmentation techniques were used to extract the slab region and background region from the above image information;

[0074] The slab region is denoised, and a minimum bounding rectangle is established based on the contour of the slab region.

[0075] It's important to note that image semantic segmentation aims to extract targets from images at the pixel level. It comprises four parts: data preprocessing, model training, training model evaluation, and new image inference. Essentially, it divides the entire image into distinct regions, assigning meaning to each. Unlike object detection, which simply locates the target in the image, boxes it, and categorizes it, semantic segmentation seamlessly divides the entire image into regions using outlines, assigning each region a category. Regions without a category are assumed to be the background.

[0076] In addition to image semantic segmentation technology, image target recognition technology can also be used to extract targets from images. The methods can be varied, such as: First, binarizing the image and distinguishing the billet region from the background region based on color difference to extract the billet region. Second, using deep learning, selecting a certain number of training and validation sets for machine learning to obtain a recognition model that can identify the billet region. The methods used in this embodiment include, but are not limited to, the above methods, as long as they achieve the recognition purpose. Establishing the minimum bounding rectangle is to facilitate the selection of marker points (such as vertices, midpoints, etc.) to locate and record the current frame position of the billet region.

[0077] Furthermore, such as Figure 6 As shown, after establishing the minimum bounding rectangle based on the contour within the aforementioned slab region, the method includes:

[0078] Step S201: When the area of ​​the slab region accounts for less than a preset threshold in the minimum bounding rectangle;

[0079] It should be noted that the slabs photographed may sometimes have defects. The main causes of these defects are object occlusion or visual analysis anomalies. Object occlusion may include: obstruction by production line components, water mist, etc. The reduction in the recognition area after occlusion can be judged by using the numerical value of the occlusion area as a threshold, thereby determining whether repair is necessary.

[0080] Step S202: Extract multiple straight line segments with a preset length from the outline of the slab area;

[0081] It should be noted that most defects in the slab image are only localized, and their shapes generally conform to rectangular characteristics. Therefore, multiple straight line segments longer than a preset length are extracted from the contour, such as... Figure 6As shown; since there are too many lines on the outline of the slab, some small defects can be ignored. Therefore, a preset length is set as the threshold for filtering, and the specific value is set according to the size of the slab.

[0082] Step S203: Select two lines from the plurality of lines and arrange them in combination, and record the two target line segments whose included angle reaches a preset angle range in each arrangement.

[0083] It should be noted that the restoration of the original area of ​​the slab is mainly based on the characteristics of a rectangle. The purpose of selecting and arranging two straight line segments is to determine whether the included angle formed by the two sides is close to a right angle or parallel, such as... Figure 7 As shown. However, in actual operation, the angle cannot be perfectly right angle or parallel, so there is a certain error. For example, the preset angle range is 85 degrees to 95 degrees.

[0084] Step S204: When the preset angle meets the first specific angle range, two corresponding straight line segments are defined to be perpendicular to each other, and the two target straight line segments are extended on the slab area image. The intersection of the extensions is taken as the vertex of the actual slab contour. When the preset angle meets the second specific angle range, two corresponding straight line segments are defined to be parallel or collinear to each other, and the two straight line segments are taken as the candidate contour.

[0085] It should be noted that the purpose of the first specific angle range is to filter line segment groups that are close to right angles. For example, the first specific angle range can be set to 85 degrees to 95 degrees. After obtaining the line segment groups that are close to right angles, the intersection point after extension is a vertex of the actual slab contour, thus obtaining the vertex position. The second specific angle range is for line segment groups that are close to parallel. For example, the second specific angle range can be set to 0 degrees to 8 degrees, thus obtaining the candidate contour based on the obtained vertex information.

[0086] Step S205: Based on the vertex position, the target line segment group, the extension length, and the candidate contour, and based on the preset features of the rectangle, repair the minimum bounding rectangle.

[0087] It should be noted that even after repairing the actual slab outline, accurate positional information still cannot be obtained. Therefore, based on the information obtained above, a new minimum bounding rectangle is constructed, i.e., the minimum bounding rectangle is repaired, and the position is obtained using the vertices and midpoints of the repaired minimum bounding rectangle. Figure 10 As shown.

[0088] Furthermore, after selecting two straight line segments as candidate contours, the method further includes:

[0089] The multiple line segment groups are arranged and combined, and the combination that meets the preset characteristics of a rectangle and contains the most line segment groups is selected. The final circumscribed rectangle is determined based on the combination.

[0090] It should be noted that the further arrangement and combination of line segment groups aims to combine line segment groups that satisfy the rectangular characteristics, thereby obtaining multiple combination methods. The group that best satisfies the rectangular characteristics is then selected as the final repair result, i.e., the group containing the largest number of line segment groups from the original image. This improves the accuracy of the repair.

[0091] Furthermore, the step of recording the position information of each frame of slab based on preset marker points on the aforementioned contour information also includes:

[0092] Using the minimum bounding rectangle as the outline, obtain the outline center of the outline;

[0093] Establish a coordinate system with the center of the above contour as the origin, and obtain the coordinates of the four vertices of the smallest bounding rectangle.

[0094] It should be noted that the position of the slab is mainly determined by the vertices and center point of the smallest bounding rectangle, such as... Figure 3 and 4 As shown, a coordinate system is established with the center of the contour as the origin, thus obtaining the coordinates of the four vertices. From this, the rotation direction and rotation angle can be calculated based on the positional changes in different frames of the image.

[0095] Furthermore, the step of outputting the rotation direction and rotation angle corresponding to the slab by comparing the positions between two adjacent frames also includes:

[0096] Based on the above sequence of multiple frames, record the initial position of the minimum bounding rectangle in the first frame;

[0097] Compare the vertex coordinates of the minimum bounding rectangle in each frame with the vertex coordinates of the previous frame, calculate the rotation angle, and determine the rotation direction based on the coordinate system.

[0098] It should be noted that although the slab is moving, the camera equipment is not moving. Therefore, the direction of the coordinate axis established by the center point of the smallest bounding rectangle will not change. However, the vertices of the smallest bounding rectangle move with the slab. Therefore, we first calculate the angle between the line formed by the vertex coordinates and the center point of each frame and the coordinate axis, and then calculate the difference of the angle to obtain the rotation angle value.

[0099] Furthermore, the step of outputting the rotation direction and rotation angle corresponding to the slab by comparing the positions between two adjacent frames further includes:

[0100] The coordinate system described above is established with the forward transport direction along the production line as the reference axis, based on the axis of symmetry and the center of the minimum bounding rectangle. The angle formed by rotating the axis of symmetry of the minimum bounding rectangle based on the reference axis is defined as the rotation angle.

[0101] If the aforementioned angle is formed in the counterclockwise rotation direction of the reference axis, the angle sign is positive; otherwise, the sign is negative.

[0102] It should be noted that the origin is the center of the smallest bounding rectangle, and the coordinate axes are the axis of symmetry of the smallest bounding rectangle. Simultaneously, the forward transport direction along the production line is used as the reference axis. This is achieved by utilizing the changing characteristics of the forward transport direction to obtain the directional reference. Since there are two axes of symmetry for the rectangle, each frame of analysis of the slab's bounding rectangle profile will result in two rotation angles, denoted as a1 and a2, as follows... Figure 4 As shown in the diagram. The defined rotation angle is the angle at which the slab rotates from its positive position. The calculation of the rotation angle difference is used for selecting candidate angles and assisting in determining the rotation direction. Specifically, rotation angles are initialized based on the first frame. The rotation angle with the smaller absolute value between a1 and a2 is selected as the rotation angle describing the initial rotation state, and its value is recorded. Starting from the second frame, the rotation angle for the current frame is determined based on the rotation angle from the previous frame, the rotation direction from the previous frame to this frame, and the analysis results of the candidate rotation angles a1 and a2. Based on the rotation direction determination result, the relationship between the rotation angle values ​​of the previous frame and the current frame is given as a constraint condition. Specifically: if the determination result indicates that the rotation from the previous frame to this frame should be counterclockwise, then the rotation angle value of this frame should be greater than the rotation angle value of the previous frame; if the determination result indicates that the rotation from the previous frame to this frame should be clockwise, then the rotation angle value of this frame should be less than the rotation angle value of the previous frame. The rotation angle between a1 and a2 that satisfies the constraint condition is selected as the final rotation angle describing the rotation state of this frame. Based on the rotation angle and constraints of the previous frame, if both a1 and a2 satisfy the constraints, the rotation angle that is smaller between the rotation angle of a1 and a2 and the rotation angle of the previous frame is taken as the rotation angle describing the rotation state of the current frame.

[0103] Furthermore, the step of extracting multiple straight line segments longer than a preset length on the outline of the slab region includes:

[0104] When there is a continuous straight line segment containing multiple slope differences within a second preset range, draw a straight line segment along the start and end points of the continuous straight line segment and extract it based on the preset length.

[0105] When a curve contains peak and valley values ​​within a third preset range, the curve is fitted into a straight line of the corresponding length using the least squares method, and the curve is extracted based on the preset length.

[0106] It should be noted that when the slab is obscured, some areas may exhibit serrations, such as... Figure 5 As shown, multiple continuous straight line segments are selected, and the slope difference is within a first preset range (this range only requires approximations of straight lines and can be freely chosen as needed). To avoid selecting lines that are too short to represent the required segments, a preset length is used for filtering and extraction. Similarly, some segments may contain jagged edges. Since curve filtering only considers curves that approximate straight lines, the peak and trough values ​​of the curve are used for filtering. After filtering, the least squares method is used to fit the curves to straight lines of corresponding lengths, and the preset length is used as a threshold for further filtering and extraction.

[0107] Furthermore, the step of extracting multiple straight line segments longer than a preset length on the outline of the slab region further includes:

[0108] When the number of straight line segments extracted from the outline of the slab area is less than two, the extraction is deemed to have failed, the current frame is abandoned, and the process jumps to the next frame to re-detect.

[0109] If multiple frames fail to be extracted, an alarm is issued indicating an anomaly in the current rotation state detection process.

[0110] It should be noted that when the shape of the slab cannot be formed, i.e., the number of straight line segments extracted from the outline of the slab area is less than two, the main reasons may be severe damage to the slab itself, severe obstruction from surrounding equipment, or excessive water mist concentration. In this case, there are no conditions for repair. Therefore, when the number of straight line segments extracted from the outline of the slab area is less than two, it indicates that the acquired image frame does not meet the requirements, and the current frame is abandoned, jumping to the next frame to begin detection. When multiple frames fail to extract the slab, it indicates a quality problem with the slab or that the surrounding environment is significantly affecting the detection, thus issuing an alarm to prompt manual intervention or other operations.

[0111] Based on the same inventive concept, embodiments of this application provide a machine vision-based system for detecting the rotational state of a slab during steel transfer, such as... Figure 11 As shown, it includes:

[0112] The video acquisition module 201 is used to acquire multi-frame image information of slabs on the steel transfer roller conveyor using a camera device;

[0113] The contour extraction module 202 is used to obtain the contour information of the slab through the above image information;

[0114] The position acquisition module 203 is used to record the position information of each frame of slab based on preset marker points on the above contour information;

[0115] The running status calculation module 204 is used to output the rotation direction and rotation angle corresponding to the slab by comparing the positions between two adjacent frames based on the sequence of the above multi-frames.

[0116] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. 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 illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0117] 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.

[0118] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0119] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for detecting the rotation state of a slab during steel transfer based on machine vision, characterized in that, The method includes: Multiple frames of images of slabs on the steel transfer roller conveyor are acquired using camera equipment; The outline information of the slab is obtained through the image information; The position information of each frame of slab is recorded based on preset marker points on the contour information; Based on the sequence of the multiple frames, the rotation direction and rotation angle corresponding to the slab are output by comparing the positions between two adjacent frames; The step of obtaining the contour information of the slab through the image information further includes: Semantic segmentation technology is used to extract the slab region and background region from the image information; Denoise the slab region and establish a minimum bounding rectangle based on the contour of the slab region; After establishing the minimum bounding rectangle based on the contour of the slab region, the method further includes: When the area of ​​the slab region accounts for less than a preset threshold in the minimum bounding rectangle, multiple straight line segments with a length greater than a preset threshold are extracted from the outline of the slab region. Select two lines from the multiple line segments and arrange them in combination. Record the two target line segments whose included angle reaches a preset angle range in each arrangement. When the preset angle range meets the first specific angle range, two corresponding straight line segments are defined to be perpendicular to each other, and the two target straight line segments are extended on the slab area image. The intersection of the extensions is taken as the vertex of the actual slab contour. When the preset angle range meets the second specific angle range, two corresponding straight line segments are defined to be parallel or collinear to each other, and the two straight line segments are taken as the candidate contour. Based on the position of the vertex, the target line segment group, the extension length, and the candidate contour, the minimum bounding rectangle is repaired based on the preset features of the rectangle; The step of outputting the rotation direction and rotation angle corresponding to the slab by comparing the positions between two adjacent frames further includes: Based on the sequence of the multiple frames, record the initial position of the minimum bounding rectangle in the first frame; The vertex coordinates of the minimum bounding rectangle in each frame are compared with the vertex coordinates of the previous frame, the rotation angle is calculated, and the rotation direction is determined according to the coordinate system.

2. The method as described in claim 1, characterized in that, After selecting two straight line segments as candidate contours, the method further includes: The target line segment groups are arranged and combined, and the combination that meets the preset characteristics of a rectangle and contains the most target line segment groups is selected. The final circumscribed rectangle is determined based on the combination.

3. The method according to any one of claims 1-2, characterized in that, The step of recording the position information of each frame of slab based on preset marker points on the contour information further includes: Using the smallest bounding rectangle as the outline, obtain the outline center of the outline; Establish a coordinate system with the center of the outline as the origin, and obtain the coordinates of the four vertices of the smallest bounding rectangle.

4. The method as described in claim 1, characterized in that, The step of extracting multiple straight line segments longer than a preset length on the contour of the slab region includes: When there is a continuous straight line segment containing multiple slope differences within a second preset range, draw a straight line segment along the start and end points of the continuous straight line segment and extract it based on the preset length; When a curve contains peak and valley values ​​within a third preset range, the curve is fitted into a straight line of corresponding length using the least squares method, and the curve is extracted based on the preset length.

5. The method as described in claim 1, characterized in that, The step of extracting multiple straight line segments longer than a preset length on the contour of the slab region further includes: When the number of straight line segments extracted from the contour of the slab area is less than two, the extraction is deemed to have failed, the current frame is discarded, and the process jumps to the next frame image to re-detect. If multiple frames fail to be extracted, an alarm is issued indicating an anomaly in the current rotation state detection process.

6. A slab rotation state detection system based on machine vision during steel transfer, characterized in that, The system includes: The video acquisition module is used to acquire multi-frame image information of slabs on the steel transfer roller conveyor using camera equipment; A contour extraction module is used to obtain the contour information of the slab through the image information; The contour extraction module is also used to extract the slab region and background region from the image information using semantic segmentation technology; Denoise the slab region and establish a minimum bounding rectangle based on the contour of the slab region; When the area of ​​the slab region accounts for less than a preset threshold in the minimum bounding rectangle, multiple straight line segments with a length greater than a preset threshold are extracted from the outline of the slab region. Select two lines from the multiple line segments and arrange them in combination. Record the two target line segments whose included angle reaches a preset angle range in each arrangement. When the preset angle range meets the first specific angle range, two corresponding straight line segments are defined to be perpendicular to each other. The two target straight line segments are extended on the slab area image, and the intersection of the extensions is taken as the vertex of the actual slab contour. When the preset angle meets the second specific angle range, two corresponding straight line segments are defined to be parallel to each other or collinear. The two straight line segments are taken as the candidate contour. Based on the position of the vertex, the target line segment group, the extension length, and the candidate contour, the minimum bounding rectangle is repaired based on the preset features of the rectangle; The position acquisition module is used to record the position information of each frame of slab based on preset marker points on the contour information; The running status calculation module is used to output the rotation direction and rotation angle corresponding to the slab by comparing the positions between two adjacent frames based on the sequence of the multiple frames; The running state calculation module is also used to record the initial position of the minimum bounding rectangle in the first frame based on the sequence of the multiple frames; The vertex coordinates of the minimum bounding rectangle in each frame are compared with the vertex coordinates of the previous frame, the rotation angle is calculated, and the rotation direction is determined according to the coordinate system.

Images