Spinneret plate cleaning method, system and equipment based on visual positioning

Abstract

A spinneret plate cleaning method, system and device based on visual localization are characterized in that an industrial camera collects an image of a spinneret plate, the image is preprocessed to obtain the preprocessed image, the preprocessed image is subjected to fitting circle calculation to obtain the circle center and radius of a hole site candidate circle, the hole site candidate circle is a spinneret plate hole possibly existing in the spinneret plate, and the number of the hole site candidate circle is larger than that of the spinneret plate hole. And round contour screening is conducted on the hole position candidate circles, the circle with the small radius is selected as a target circle, the pixel coordinates of the target circle are obtained, interference caused by an error circle with the highest roundness score is prevented, the pixel coordinates of the target circle are converted into mechanical coordinates, then a cleaning path is planned, and a mechanical arm is driven to clean the spinneret plate holes. According to the design, when contour lines around the target are relatively complex, interference of abnormal values exists and an approximate arc contour exists, interference of an error circle with the highest roundness score can be prevented, the image can be preprocessed through global threshold processing, and the requirement for the imaging quality is effectively lowered.

Description

Technical Field

[0001] The present invention relates to a spinneret cleaning method, and in particular to a spinneret cleaning method, system and equipment based on visual positioning. Background Art

[0002] The spinneret is a key component in chemical fiber production, and its cleanliness directly affects product quality and production efficiency. Over time, various dirt and residues accumulate on the spinneret, so regular cleaning is essential.

[0003] The traditional spinneret cleaning process usually relies on manual operation or simple mechanical equipment, which has problems such as low efficiency, high labor intensity, and insufficient cleaning accuracy. Especially when the spinneret surface is complex, the holes are small and densely distributed, manual cleaning is prone to omissions or damage. In addition, most existing automated cleaning equipment relies on fixed tracks or simple sensors, lacks the ability to perceive the spinneret position and cleaning status in real time, and is difficult to adapt to dynamic changes in complex industrial environments.

[0004] Therefore, the industry has proposed an intelligent cleaning system based on visual positioning to minimize manual operations during the positioning process, achieve precise positioning and efficient cleaning of the spinneret, and improve processing efficiency and accuracy. Common intelligent cleaning algorithms based on visual positioning include the Hough transform algorithm to obtain the center coordinates and radius of the circular hole, and the use of Euclidean distance transform to calculate the position of the center of the circle.

[0005] Although these spinneret cleaning methods based on visual positioning are feasible, they still have the following defects:

[0006] 1. It is easily affected by outliers and cannot eliminate the interference of approximate arc contours and complex backgrounds in close positions.

[0007] 2. The imaging quality of the edge of the part to be measured is required to be high, and it is easily affected by edge position defects, resulting in inaccurate center positioning.

[0008] The information disclosed in this background technology section is only intended to increase understanding of the overall background of the application and should not be considered as an admission or any form of suggestion that the information constitutes the prior art already known to ordinary technicians in this field. Summary of the Invention

[0009] The purpose of the present invention is to overcome the shortcomings of the prior art that it is easily disturbed by abnormal values ​​and has high requirements for imaging quality, and to provide a spinneret cleaning method, system and equipment based on visual positioning that is not easily disturbed by abnormal values ​​and has low requirements for imaging quality.

[0010] To achieve the above objectives, the technical solution of the present invention is:

[0011] A spinneret cleaning method based on visual positioning, the method of use is based on the following structure:

[0012] Image acquisition system, motion control system, clamping system and industrial computer;

[0013] The image acquisition system is an industrial camera with a ring light source. The image acquisition system is placed directly opposite the spinneret that needs to be cleaned. The image acquisition system is used to acquire images of the spinneret and transmit the image information to an industrial computer for processing. The motion control system is a robotic arm. The end of the motion control system is fixedly connected to a clamping system. The clamping system is used to clamp the cleaning laser head. The industrial computer is electrically connected to the image acquisition system, the motion control system, and the clamping system respectively. The industrial computer is used to control the operation of the image acquisition system, the motion control system, and the clamping system.

[0014] The method of use comprises the following steps:

[0015] S1. Capturing a spinneret image through an industrial camera and preprocessing the image to obtain a preprocessed image;

[0016] S2. Performing circle fitting calculation on the pre-processed image to obtain the center and radius of a hole position candidate circle, where the hole position candidate circle is a spinneret hole that may exist on the spinneret;

[0017] S3, performing circular contour screening on the hole position candidate circles to obtain the pixel coordinates of the target circle, where the target circle is the spinneret hole that needs to be cleaned;

[0018] S4. Convert the pixel coordinates of the target circle into mechanical coordinates, where the origin of the coordinate system of the mechanical coordinates is the connection point between the robot arm and the base in the motion control system.

[0019] In S1, the image preprocessing includes using a global threshold processing method, performing opening and closing operations to obtain the area where the circular hole in the image is located as the region of interest, and setting the side length of the region of interest to no more than 3 times the spinneret aperture;

[0020] The image of the region of interest is enhanced by using denoising, filtering and smoothing, and contrast enhancement algorithms to obtain a preprocessed image.

[0021] In S2, the fitting circle calculation includes setting the search range of the caliper tool, detecting the area where the target edge may appear within the search range by the caliper tool, setting the number of search edges, obtaining all edge point sets of each caliper detection output result, performing a fitting circle operation on all edge point sets, and obtaining the center and radius of the hole candidate circle.

[0022] In S2, the operation of setting the search range of the caliper tool includes:

[0023] Use the area_center() operator to get the center point of the region of interest. The area_center() operator performs a binarization of the connected regions in the image, scans all pixels in the region, counts the total number of pixels as the area of ​​the region, sums the horizontal and vertical coordinates of all pixels, and divides the sum by the area to get the center coordinates. The center point of the region of interest is used as the center point of the caliper measurement tool.

[0024] Use the add_metrology_object_circle_measure() operator to set the measurement range of the caliper tool according to the nominal radius of the spinneret hole and the actual machining error.

[0025] The operations for setting the number of edges to be found and obtaining the set of all edge points in the output results of each caliper detection include:

[0026] Use the operator set_metrology_object_param(MetrologyHandle,CircleIndex,'num_instances',2) to set the number of circles to 2, that is, the measuring caliper detects and outputs two circles, and use get_metrology_object_measures(Contours,

[0027] The MetrologyHandle,'all','all',Rows,Columns) operator generates multiple caliper tools along the preset normal. Each caliper locates the local maximum gradient point through edge gradient analysis and retains the points that meet the contrast threshold and edge polarity to obtain the set of all edge points, namely (Rows, Columns).

[0028] The operation of fitting a circle to all edge point sets includes:

[0029] Use the fit_circle_contour_xld(Contour,'algebraic',-1,0,0,3,2,RowCircle,ColumnCircle,RadiusCircle,StartPhi,EndPhi,PointOrder) operator to approximate the arc and obtain the center point coordinates (RowCircle, ColumnCircle) and radius RadiusCircle data of the circle;

[0030] Use the circle equation to approximate a curve by an arc:

[0031] x 2 -2Ax+y 2-2By-C=0;

[0032] In the above formula, A represents the RowCircle calculated by the fit_circle_contour_xld operator in step S4, which is used to represent the abscissa of the center of the spinneret hole. B represents the ColumnCircle calculated by the fit_circle_contour_xld operator in step S4, which is used to represent the ordinate of the center of the spinneret hole. C represents the constant term related to the radius. x represents the abscissa of the edge point, and y represents the ordinate of the edge point.

[0033] The error of point i on the curve relative to the arc includes:

[0034] δ i =z i -2Ax i -2By i -C;

[0035] In the above formula, Used to simplify the algebraic expression of the circle equation, xi represents the horizontal coordinate of the i-th point, yi represents the vertical coordinate of the i-th point, δ i Represents the degree to which the i-th point deviates from the fitting circle according to the algebraic distance;

[0036] Minimize the sum of squares of the errors of the approximating arc for each point on the circle to ensure that the arc approaches the circle infinitely:

[0037]

[0038] By calculating the three constants A, B, and C through extreme values ​​and elimination, we can get the coordinates of the center of the circle (A, B) and the radius.

[0039] In S3, the circle with the smaller radius among the two hole position candidate circles is used as the target circle, and the sequence number Min of the target circle is obtained. The specific operations include:

[0040] Use the gen_circle_contour_xld() operator to calculate the contour ContCircleMin of the target circle, and then use the div_display() operator to display it. By observing the deviation between the contour edge and the actual target contour, verify whether it is the final target contour;

[0041] The coordinates of the target center point are (RowCircle[Min], ColumnCircle[Min]), and the coordinates at this time are the pixel coordinates in the image;

[0042] Taking the intersection of the robotic arm and the chassis as the origin, use the image_points_to_world_plane() operator to convert the pixel coordinates of the center point into the world coordinate system. Use the hand-eye calibration matrix and the affine_trans_point_3d() operator and the matrix transformation operator to calculate the coordinates of the center point of the circular component in the robotic arm base coordinate system.

[0043] In S3, the operation of observing the deviation between the contour edge and the actual target contour includes:

[0044] Use distance_pc() to calculate the distance from each contour point to the center of the fitting circle and store it in the array Distance; calculate the residual root mean square as the roughness index, the formula is:

[0045]

[0046] In the above formula, N is the number of contour points, Distance[i] is the distance from each contour point to the center of the fitting circle, and RadiusCircle is the radius of the fitting circle;

[0047] If the threshold Roughness_Index>0.8μm, it indicates that there is local corrosion on the profile;

[0048] Use fit_ellipse_contour_xld() to fit an ellipse to the same contour and obtain the major axis (MajorAxis) and minor axis (MinorAxis); then calculate the ellipse dispersion index, the formula is:

[0049]

[0050] If the threshold Ellipticity_Error>0.05, it indicates that the contour may be deformed or interfered by non-concentric ellipses. The weighted least squares method is used to refit the ellipse to suppress the influence of outliers.

[0051] Therefore, by observing the deviation between the contour edge and the actual target contour, it is verified whether it is the final target contour.

[0052] In S4, the operation of converting the pixel coordinates of the target circle into mechanical coordinates by the robotic arm includes:

[0053] The coordinates of the target center point are (RowCircle[Min], ColumnCircle[Min]), and the coordinates at this time are the pixel coordinates in the image;

[0054] Use the image_points_to_world_plane() operator to convert the pixel coordinates of the center point into the world coordinate system;

[0055] Taking the intersection of the robot arm and the chassis as the origin of the coordinate system, the hand-eye calibration matrix, the affine_trans_point_3d() operator, and the matrix transformation operator are used to calculate the coordinates of the center point of the circular component in the robot base coordinate system.

[0056] A spinneret cleaning system based on visual positioning, the system is used to perform the aforementioned spinneret cleaning method based on visual positioning, specifically comprising: an image acquisition module, a preprocessing module, a hole position screening module, and a cleaning module;

[0057] The image acquisition module is used to: acquire the spinneret image through an industrial camera, and pre-process the image to obtain a pre-processed image;

[0058] The preprocessing module is used to: perform fitting circle calculation on the preprocessed image to obtain the center and radius of the hole position candidate circle, where the hole position candidate circle is the spinneret hole that may exist on the spinneret;

[0059] The hole position screening module is used to: perform circular contour screening on the hole position candidate circles to obtain the pixel coordinates of the target circle, and the target circle is the spinneret hole that needs to be cleaned;

[0060] The cleaning module is used to convert the pixel coordinates of the target circle into mechanical coordinates, where the origin of the coordinate system of the mechanical coordinates is the connection part between the mechanical arm and the base in the motion control system.

[0061] A spinneret cleaning device based on visual positioning comprises a memory and a processor, wherein the memory is used to store computer program code and transmit the computer program code to the processor;

[0062] The processor is configured to execute the aforementioned spinneret cleaning method based on vision positioning according to instructions in the computer program code.

[0063] A computer program product includes a computer program, wherein the computer program is executed by a processor to execute the aforementioned spinneret cleaning method based on vision positioning.

[0064] Compared with the prior art, the present invention has the following beneficial effects:

[0065] 1. In a spinneret cleaning method based on visual positioning of the present invention, an industrial camera is used to collect spinneret images, and the images are preprocessed to obtain a preprocessed image. A circle fitting calculation is performed on the preprocessed image to obtain the center and radius of the hole position candidate circle. The hole position candidate circle is a spinneret hole that may exist on the spinneret. The hole position candidate circle is screened for a circular contour to obtain the pixel coordinates of the target circle. The pixel coordinates of the target circle are converted into mechanical coordinates, and then a cleaning path is planned to drive the mechanical arm to clean the spinneret hole. Therefore, this design can realize the fully automatic cleaning of the spinneret, effectively improving the degree of automation.

[0066] 2. In the spinneret cleaning method based on visual positioning of the present invention, when calculating the fitted circle, a caliper tool is used to find two candidate circles for the hole position, and the circle with the smaller radius is selected as the target circle. This allows the method to prevent interference from the erroneous circle with the highest roundness score when the contour lines around the target are relatively complex, there are interference from outliers, and there are approximate circular arc contours. Therefore, this design can prevent outliers from interfering with visual positioning.

[0067] 3. In the spinneret cleaning method based on visual positioning of the present invention, global threshold processing is used. After opening and closing operations, the area where the circular hole in the image is located is obtained as the region of interest. The image of the region of interest is then enhanced using denoising, filtering and smoothing, and contrast enhancement algorithms to obtain the main area. The measurement range of the caliper tool is set according to the nominal radius of the spinneret hole and the actual processing error. The caliper tool is used to find the edge point set of the main area, and then the hole position candidate circle is obtained. The imaging quality requirements for the edge of the part to be tested are relatively low. Therefore, this design can pre-process the image through global threshold processing, effectively reducing the requirements for imaging quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 is a flow chart of the method of the present invention.

[0069] Figure 2 It is a logic flow chart of the algorithm of the present invention.

[0070] Figure 3 It is a structural diagram of the system of the present invention.

[0071] Figure 4 This is a structural diagram of the device described in Example 3. DETAILED DESCRIPTION

[0072] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0073] Example 1:

[0074] See also Figure 1, a spinneret cleaning method based on visual positioning, the method of use is based on the following structure:

[0075] Image acquisition system, motion control system, clamping system and industrial computer;

[0076] The image acquisition system is an industrial camera with a ring light source. The image acquisition system is placed directly opposite the spinneret that needs to be cleaned. The image acquisition system is used to acquire images of the spinneret and transmit the image information to an industrial computer for processing. The motion control system is a robotic arm. The end of the motion control system is fixedly connected to a clamping system. The clamping system is used to clamp the cleaning laser head. The industrial computer is electrically connected to the image acquisition system, the motion control system, and the clamping system respectively. The industrial computer is used to control the operation of the image acquisition system, the motion control system, and the clamping system.

[0077] The method of use comprises the following steps:

[0078] S1. Capturing a spinneret image through an industrial camera and preprocessing the image to obtain a preprocessed image;

[0079] S2. Performing circle fitting calculation on the pre-processed image to obtain the center and radius of a hole position candidate circle, where the hole position candidate circle is a spinneret hole that may exist on the spinneret;

[0080] S3, performing circular contour screening on the hole position candidate circles to obtain the pixel coordinates of the target circle, where the target circle is the spinneret hole that needs to be cleaned;

[0081] S4. Convert the pixel coordinates of the target circle into mechanical coordinates, where the origin of the coordinate system of the mechanical coordinates is the connection point between the robot arm and the base in the motion control system.

[0082] See also Figure 2 In S1, the image preprocessing includes using a global threshold processing method, performing opening and closing operations to obtain the area where the circular hole in the image is located as the region of interest, and setting the side length of the region of interest to no more than 3 times the spinneret aperture, generally 3 mm;

[0083] The image of the region of interest is enhanced by using denoising, filtering and smoothing, and contrast enhancement algorithms to obtain a preprocessed image.

[0084] In S2, the fitting circle calculation includes setting the search range of the caliper tool, detecting the area where the target edge may appear within the search range by the caliper tool, setting the number of search edges, obtaining all edge point sets of each caliper detection output result, performing a fitting circle operation on all edge point sets, and obtaining the center and radius of the hole candidate circle.

[0085] In S2, the operation of setting the search range of the caliper tool includes:

[0086] Use the area_center() operator to get the center point of the region of interest. The area_center() operator performs a binarization of the connected regions in the image, scans all pixels in the region, counts the total number of pixels as the area of ​​the region, sums the horizontal and vertical coordinates of all pixels, and divides the sum by the area to get the center coordinates. The center point of the region of interest is used as the center point of the caliper measurement tool.

[0087] Use the add_metrology_object_circle_measure() operator to set the measurement range of the caliper tool according to the nominal radius of the spinneret hole and the actual machining error.

[0088] The operations for setting the number of edges to be found and obtaining the set of all edge points in the output results of each caliper detection include:

[0089] Use the set_metrology_object_param(MetrologyHandle,CircleIndex,'num_instances',2) operator to set the number of circles to be found to 2, that is, the measuring caliper detects and outputs two circles, and uses the get_metrology_object_measures(Contours,MetrologyHandle,'all','all',Rows,Columns) operator to generate multiple caliper tools along the preset normal. Each caliper locates the local maximum gradient point through edge gradient analysis, and retains the points that meet the contrast threshold and edge polarity to obtain the set of all edge points, that is, (Rows, Columns).

[0090] The fit_circle_contour_xld(Contour,'algebraic',-1,0,0,3,2,RowCircleColumnCircle,RadiusCircle,StartPhi,EndPhi,PointOrder) operator is used to approximate the arc and obtain the center point coordinates (RowCircle,ColumnCircle) and radius RadiusCircle data of the circle. The operator fit_circle_contour_xld() supports six circle fitting methods: algebraic, ahuber, atukey, geometric, geohuber, and geotukey. A total of 200 images were collected during the experiment, and overexposure, occlusion, and corrosion-type failure images were detected in turn through a multi-level dynamic judgment mechanism. First, laser overexposure images were identified based on the grayscale saturation ratio of the region of interest (>15%) and the maximum grayscale threshold (≥250). Next, dirt obstruction was determined by combining the edge library fracture degree (>0.25) and the gradient direction standard deviation (>35°). Finally, oxidation corrosion failure was detected using the radial residual root mean square error (>0.8μm) and the ellipse fitting axis ratio deviation (>0.05). A priority series judgment strategy was used to eliminate 40 images with positioning failures caused by laser spot overexposure, residual dirt obstruction, and severe micropore oxidation corrosion. The remaining 160 images were then subjected to sub-pixel fitting calculation and analysis of the micropore center point.

[0091] The operation of fitting a circle to all edge point sets includes:

[0092] Use the circle equation to approximate a curve by an arc:

[0093] x 2 -2Ax+y 2 -2By-C=0;

[0094] In the above formula, A represents the RowCircle calculated by the fit_circle_contour_xld operator in step S4, which is used to represent the abscissa of the center of the spinneret hole. B represents the ColumnCircle calculated by the fit_circle_contour_xld operator in step S4, which is used to represent the ordinate of the center of the spinneret hole. C represents the constant term related to the radius. x represents the abscissa of the edge point, and y represents the ordinate of the edge point.

[0095] The error of point i on the curve relative to the arc includes:

[0096] δ i =z i -2Ax i -2By i-C;

[0097] In the above formula, Used to simplify the algebraic expression of the circle equation, xi represents the horizontal coordinate of the i-th point, yi represents the vertical coordinate of the i-th point, δ i Represents the degree to which the i-th point deviates from the fitting circle according to the algebraic distance;

[0098] Minimize the sum of squares of the errors of the approximating arc for each point on the circle to ensure that the arc approaches the circle infinitely:

[0099]

[0100] By calculating the three constants A, B, and C through extreme values ​​and elimination, we can get the coordinates of the center of the circle (A, B) and the radius.

[0101] In the above S3, the contour lines around the detection image are relatively complex, and the target circle is easily affected by other circular contours. If only one circle is output, the circle with the highest roundness score will be detected in the circle set of the detection results, and the result is uncertain. After analysis, the position where false detection is prone to occur is the circumscribed circle closest to the target contour. To solve this problem, the purpose of this step is to find two circles, and by judging the radius value of the circle, filter out the circle with the smaller radius of the two circles and obtain its serial number Min. If the detection result is that only one circle is found, then this circle is defaulted to be the target circle and is directly used in subsequent calculations. The circle with the smaller radius of the two hole position candidate circles is used as the target circle, and the serial number Min of the target circle is obtained. The specific operations include:

[0102] Use the gen_circle_contour_xld() operator to calculate the contour ContCircleMin of the target circle, and then use the div_display() operator to display it. By observing the deviation between the contour edge and the actual target contour, verify whether it is the final target contour;

[0103] The coordinates of the target center point are (RowCircle[Min], ColumnCircle[Min]), and the coordinates at this time are the pixel coordinates in the image;

[0104] Taking the intersection of the robotic arm and the chassis as the origin, use the image_points_to_world_plane() operator to convert the pixel coordinates of the center point into the world coordinate system. Use the hand-eye calibration matrix and the affine_trans_point_3d() operator and the matrix transformation operator to calculate the coordinates of the center point of the circular component in the robotic arm base coordinate system.

[0105] In S3, the operation of observing the deviation between the contour edge and the actual target contour includes:

[0106] Use distance_pc() to calculate the distance from each contour point to the center of the fitting circle and store it in the array Distance; calculate the residual root mean square as the roughness index, the formula is:

[0107]

[0108] In the above formula, N is the number of contour points, Distance[i] is the distance from each contour point to the center of the fitting circle, and RadiusCircle is the radius of the fitting circle;

[0109] If the threshold Roughness_Index>0.8μm, it indicates that there is local corrosion on the profile;

[0110] Use fit_ellipse_contour_xld() to fit an ellipse to the same contour and obtain the major axis (MajorAxis) and minor axis (MinorAxis); then calculate the ellipse dispersion index, the formula is:

[0111]

[0112] If the threshold Ellipticity_Error>0.05, it indicates that the contour may be deformed or interfered by non-concentric ellipses. The weighted least squares method (weight = 1 / (distance mean)^2) is used to refit the ellipse to suppress the influence of outliers.

[0113] Therefore, by observing the deviation between the contour edge and the actual target contour, it is verified whether it is the final target contour.

[0114] In S4, the operation of converting the pixel coordinates of the target circle into mechanical coordinates by the robotic arm includes:

[0115] The coordinates of the target center point are (RowCircle[Min], ColumnCircle[Min]), and the coordinates at this time are the pixel coordinates in the image;

[0116] Use the image_points_to_world_plane() operator to convert the pixel coordinates of the center point into the world coordinate system;

[0117] Taking the intersection of the robot arm and the chassis as the origin of the coordinate system, the hand-eye calibration matrix, the affine_trans_point_3d() operator, and the matrix transformation operator are used to calculate the coordinates of the center point of the circular component in the robot base coordinate system.

[0118] Example 2:

[0119] See also Figure 3, a spinneret cleaning system based on visual positioning, the system is used to perform the spinneret cleaning method based on visual positioning as described in Example 1, specifically comprising: an image acquisition module, a preprocessing module, a hole position screening module, and a cleaning module;

[0120] The image acquisition module is used to: acquire the spinneret image through an industrial camera, and pre-process the image to obtain a pre-processed image;

[0121] The preprocessing module is used to: perform fitting circle calculation on the preprocessed image to obtain the center and radius of the hole position candidate circle, where the hole position candidate circle is the spinneret hole that may exist on the spinneret;

[0122] The hole position screening module is used to: perform circular contour screening on the hole position candidate circles to obtain the pixel coordinates of the target circle, and the target circle is the spinneret hole that needs to be cleaned;

[0123] The cleaning module is used to convert the pixel coordinates of the target circle into mechanical coordinates, where the origin of the coordinate system of the mechanical coordinates is the connection part between the mechanical arm and the base in the motion control system.

[0124] Example 3:

[0125] See also Figure 4 ,A spinneret cleaning device based on vision positioning, comprising a memory and a processor, wherein the memory is used to store computer program code and transmit the computer program code to the processor;

[0126] The processor is used to execute the spinneret cleaning method based on visual positioning as described in Example 1 according to the instructions in the computer program code.

[0127] A computer program product includes a computer program, which is executed by a processor to implement the spinneret cleaning method based on vision positioning as described in Example 1.

[0128] The above description is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiment. Any equivalent modifications or changes made by ordinary technicians in this field based on the contents disclosed in the present invention should be included in the protection scope recorded in the claims.

Claims

1. A spinneret cleaning method based on visual positioning, characterized in that: The usage is based on the following structure: Image acquisition system, motion control system, clamping system and industrial computer; The image acquisition system is an industrial camera with a ring light source. The image acquisition system is placed directly opposite the spinneret that needs to be cleaned. The image acquisition system is used to acquire images of the spinneret and transmit the image information to an industrial computer for processing. The motion control system is a robotic arm. The end of the motion control system is fixedly connected to a clamping system. The clamping system is used to clamp the cleaning laser head. The industrial computer is electrically connected to the image acquisition system, the motion control system, and the clamping system respectively. The industrial computer is used to control the operation of the image acquisition system, the motion control system, and the clamping system. The method of use comprises the following steps: S1. Capturing a spinneret image through an industrial camera and preprocessing the image to obtain a preprocessed image; S2. Performing circle fitting calculation on the pre-processed image to obtain the center and radius of a hole position candidate circle, where the hole position candidate circle is a spinneret hole that may exist on the spinneret; S3, performing circular contour screening on the hole position candidate circles to obtain the pixel coordinates of the target circle, where the target circle is the spinneret hole that needs to be cleaned; S4. Convert the pixel coordinates of the target circle into mechanical coordinates, where the origin of the coordinate system of the mechanical coordinates is the connection point between the robot arm and the base in the motion control system.

2. The spinneret cleaning method based on visual positioning according to claim 1, characterized in that: In S1, the image preprocessing includes using a global threshold processing method, performing opening and closing operations to obtain the area where the circular hole in the image is located as the region of interest, and setting the side length of the region of interest to no more than 3 times the spinneret aperture; The image of the region of interest is enhanced by using denoising, filtering and smoothing, and contrast enhancement algorithms to obtain a preprocessed image.

3. The spinneret cleaning method based on visual positioning according to claim 1, characterized in that: In S2, the fitting circle calculation includes setting the search range of the caliper tool, detecting the area where the target edge may appear within the search range by the caliper tool, setting the number of search edges, obtaining all edge point sets of each caliper detection output result, performing a fitting circle operation on all edge point sets, and obtaining the center and radius of the hole candidate circle.

4. The spinneret cleaning method based on visual positioning according to claim 3, characterized in that: In S2, the operation of setting the search range of the caliper tool includes: Use the area_center() operator to get the center point of the region of interest. The area_center() operator performs a binarization of the connected regions in the image, scans all pixels in the region, counts the total number of pixels as the area of ​​the region, sums the horizontal and vertical coordinates of all pixels, and divides the sum by the area to get the center coordinates. The center point of the region of interest is used as the center point of the caliper measurement tool. Use the add_metrology_object_circle_measure() operator to set the measurement range of the caliper tool according to the nominal radius of the spinneret hole and the actual machining error. The operations for setting the number of edges to be found and obtaining the set of all edge points of each caliper detection output result include: Use the set_metrology_object_param(MetrologyHandle,CircleIndex,'num_instances',2) operator to set the number of circles to be found to 2, that is, the measuring caliper detects and outputs two circles, and uses the get_metrology_object_measures(Contours,MetrologyHandle,'all','all',Rows,Columns) operator to generate multiple caliper tools along the preset normal. Each caliper locates the local maximum gradient point through edge gradient analysis, and retains the points that meet the contrast threshold and edge polarity to obtain the set of all edge points, that is, (Rows, Columns). The operation of fitting a circle to all edge point sets includes: Use the fit_circle_contour_xld(Contour,'algebraic',-1,0,0,3,2,RowCircle,ColumnCircle,RadiusCircle,StartPhi,EndPhi,PointOrder) operator to approximate the arc and obtain the center point coordinates (RowCircle, ColumnCircle) and radius RadiusCircle data of the circle; Use the circle equation to approximate a curve by an arc: x 2 -2Ax+y 2 -2By-C=0; In the above formula, A represents the RowCircle calculated by the fit_circle_contour_xld operator in step S4, which is used to represent the abscissa of the center of the spinneret hole. B represents the ColumnCircle calculated by the fit_circle_contour_xld operator in step S4, which is used to represent the ordinate of the center of the spinneret hole. C represents the constant term related to the radius. x represents the abscissa of the edge point, and y represents the ordinate of the edge point. The error of point i on the curve relative to the arc includes: δ i =z i -2Ax i -2By i -C; In the above formula, Used to simplify the algebraic expression of the circle equation, xi represents the horizontal coordinate of the i-th point, yi represents the vertical coordinate of the i-th point, δ i Represents the degree to which the i-th point deviates from the fitting circle according to the algebraic distance; Minimize the sum of squares of the errors of the approximating arc for each point on the circle to ensure that the arc approaches the circle infinitely: By calculating the three constants A, B, and C through extreme values ​​and elimination, we can get the coordinates of the center of the circle (A, B) and the radius.

5. The spinneret cleaning method based on visual positioning according to claim 1, characterized in that: In S3, the circle with the smaller radius among the two hole position candidate circles is used as the target circle, and the sequence number Min of the target circle is obtained. The specific operations include: Use the gen_circle_contour_xld() operator to calculate the contour ContCircleMin of the target circle, and then use the div_display() operator to display it. By observing the deviation between the contour edge and the actual target contour, verify whether it is the final target contour; The coordinates of the target center point are (RowCircle[Min], ColumnCircle[Min]), and the coordinates at this time are the pixel coordinates in the image; Taking the intersection of the robotic arm and the chassis as the origin, use the image_points_to_world_plane() operator to convert the pixel coordinates of the center point into the world coordinate system. Use the hand-eye calibration matrix and the affine_trans_point_3d() operator and the matrix transformation operator to calculate the coordinates of the center point of the circular component in the robotic arm base coordinate system.

6. The spinneret cleaning method based on visual positioning according to claim 5, characterized in that: In S3, the operation of observing the deviation between the contour edge and the actual target contour includes: Use distance_pc() to calculate the distance from each contour point to the center of the fitting circle and store it in the array Distance; calculate the residual root mean square as the roughness index, the formula is: In the above formula, N is the number of contour points, Distance[i] is the distance from each contour point to the center of the fitting circle, and RadiusCircle is the radius of the fitting circle; If the threshold Roughness_Index>0.8μm, it indicates that there is local corrosion on the profile; Use fit_ellipse_contour_xld() to fit the same contour to obtain the major axis MajorAxis and the minor axis MinorAxis; then calculate the ellipse dispersion index, the formula is: If the threshold Ellipticity_Error>0.05, it indicates that the contour may be deformed or interfered by non-concentric ellipses. The weighted least squares method is used to refit the ellipse to suppress the influence of outliers. Therefore, by observing the deviation between the contour edge and the actual target contour, it is verified whether it is the final target contour.

7. The spinneret cleaning method based on visual positioning according to claim 1, characterized in that: In S4, the operation of converting the pixel coordinates of the target circle into mechanical coordinates by the robotic arm includes: The coordinates of the target center point are (RowCircle[Min], ColumnCircle[Min]), and the coordinates at this time are the pixel coordinates in the image; Use the image_points_to_world_plane() operator to convert the pixel coordinates of the center point into the world coordinate system; Taking the intersection of the robot arm and the chassis as the origin of the coordinate system, the hand-eye calibration matrix, the affine_trans_point_3d() operator, and the matrix transformation operator are used to calculate the coordinates of the center point of the circular component in the robot base coordinate system.

8. A spinneret cleaning system based on visual positioning, characterized in that: The system is used to perform the spinneret cleaning method based on visual positioning according to any one of claims 1 to 7, specifically comprising: an image acquisition module, a preprocessing module, a hole position screening module, and a cleaning module; The image acquisition module is used to: acquire the spinneret image through an industrial camera, and pre-process the image to obtain a pre-processed image; The preprocessing module is used to: perform fitting circle calculation on the preprocessed image to obtain the center and radius of the hole position candidate circle, where the hole position candidate circle is the spinneret hole that may exist on the spinneret; The hole position screening module is used to: perform circular contour screening on the hole position candidate circles to obtain the pixel coordinates of the target circle, and the target circle is the spinneret hole that needs to be cleaned; The cleaning module is used to convert the pixel coordinates of the target circle into mechanical coordinates, where the origin of the coordinate system of the mechanical coordinates is the connection part between the mechanical arm and the base in the motion control system.

9. A spinneret cleaning device based on visual positioning, characterized in that: comprising a memory and a processor, wherein the memory is configured to store computer program code and transmit the computer program code to the processor; The processor is configured to execute the spinneret cleaning method based on vision positioning according to any one of claims 1 to 7 according to instructions in the computer program code.

10. A computer program product comprising a computer program, characterized in that The computer program is executed by a processor to implement the spinneret cleaning method based on vision positioning according to any one of claims 1 to 7.

Images