Anode slime cleaning path planning method and device, equipment and storage medium

Abstract

The application discloses an anode slime cleaning path planning method, device and equipment and a storage medium, and relates to the technical field of image processing. The method comprises the following steps: calculating a global tilt angle according to the circumscribed rectangle parameters corresponding to each hole connected domain in an anode slime original image, and performing image geometric correction according to the global tilt angle; extracting a global row distance parameter from the corrected image to extract a hole row center line from the corrected image; calculating a hole average height parameter according to the circumscribed rectangle parameters corresponding to the hole connected domain, and further determining a work track coordinate set; performing single-threshold segmentation on a single-channel gray image corresponding to the corrected image to obtain a candidate anode slime binary mask; performing physical appearance constraint on a plurality of anode slime connected domains in the candidate anode slime binary mask to obtain a target anode slime binary mask, and then generating a cleaning track coordinate. In this way, the cleaning work track with zero collision risk can be safely deduced under the condition that the edge hole features are incomplete.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and in particular to a method, apparatus, equipment and storage medium for planning anode mud cleaning path. Background Technology

[0002] The array of circular holes on the anode plate surface not only serves as channels for electrolyte circulation but also constitutes a physical no-go zone that the cutting head must avoid during cleaning operations. In an "eye-in-hand" camera mounting configuration, the holes at the top and bottom edges of the image are often presented as incomplete arcs due to the limited effective field of view (FOV). Conventional circle detection algorithms based on complete geometric features are prone to failure when facing such edge blind spots, preventing the system from generating safe obstacle avoidance boundaries. This can lead to engineering accidents such as the robotic arm cutting head cutting into the holes, causing equipment chipping, or electrode plate deformation. Summary of the Invention

[0003] In view of this, the purpose of the present invention is to overcome the shortcomings of the prior art and provide a method, apparatus, equipment and storage medium for planning the path of anode mud cleaning.

[0004] This invention provides the following technical solution: In a first aspect, the present invention provides a method for planning anode mud cleaning paths, the method comprising: Acquire the original image of the anode mud in the target area, and extract multiple connected domains of pores and their corresponding bounding rectangle parameters from the original image of the anode mud. The global tilt angle is calculated based on the circumscribed rectangle parameters corresponding to each of the hole connected regions. The original image of the anode mud is then geometrically corrected based on the global tilt angle to obtain the corrected image. The global row spacing parameter is obtained based on the corrected image. The center lines of all holes in the corrected image are extracted based on the global row spacing parameter. The average height parameter of the hole is calculated based on the bounding rectangle parameter corresponding to each hole's connected region. The coordinate set of the work track is determined based on the global row spacing parameter, the center lines of all holes, and the average height parameter of the hole. A single-channel grayscale image is obtained based on the corrected image, and single-threshold segmentation is performed based on the single-channel grayscale image to obtain a candidate anode mud binary mask; The target anode mud binary mask is obtained by physically constraining multiple connected domains of anode mud in the candidate anode mud binary mask; The cleaning trajectory coordinates are generated based on the set of work trajectory coordinates and the binary mask of the target anode mud.

[0005] In an optional implementation, the step of physically constraining the connected regions of each anode slime in the candidate anode slime binary mask to obtain the target anode slime binary mask includes: Calculate the pixel area of ​​each connected region of anode mud in the candidate anode mud binary mask, and filter the connected regions of anode mud according to the preset area threshold and the pixel area of ​​each connected region of anode mud to obtain the first filtered binary mask. Calculate the local variance of gray values ​​of each anode mud connected domain in the first filtered binary mask, and perform anode mud connected domain filtering according to the preset roughness threshold and the local variance of gray values ​​of each anode mud connected domain to obtain the second filtered binary mask. Calculate the density ratio of the number of pixels at the edge of each connected domain of the anode mud in the second filtered binary mask to the density ratio of the total pixel area inside the connected domain. Based on the preset complexity threshold and the density ratio of each connected domain of the anode mud, filter the connected domains of the anode mud to obtain the target anode mud binary mask.

[0006] In an optional implementation, generating the cleaning trajectory coordinates based on the set of work trajectory coordinates and the target anode mud binary mask includes: For each work track in the set of work track coordinates, a strip region with a preset longitudinal height is extracted along the work track; The mask region corresponding to the strip region is determined based on the binary mask of the target anode mud, and the pixel density distribution is obtained by performing column direction integration on the mask region. The pixel density distribution is traversed according to a preset start and end density threshold to obtain the target segment corresponding to the work track. The target segment is restored into a two-dimensional rectangular bounding box as the cleaning rectangular area based on the preset rectangular horizontal expansion compensation pixels and the preset vertical height. The cleaning trajectory coordinates corresponding to the operation track are then obtained by mapping the cleaning rectangular area.

[0007] In an optional implementation, the preset start and end density thresholds include a preset start density threshold and a preset end density threshold. The step of traversing the pixel density distribution according to the preset start and end density thresholds to obtain the target segment corresponding to the work track includes: When the pixel density distribution is traversed until the current pixel density is greater than the preset starting density threshold, the pixel corresponding to the current pixel density is taken as the starting point of the target segment. When the pixel density distribution is traversed until the current pixel density is less than the preset termination density threshold, and the state where the current pixel density is less than the preset termination density threshold exceeds a preset distance, then the pixel point corresponding to the current pixel density distribution is taken as the endpoint of the target segment.

[0008] In an optional implementation, obtaining the cleaning trajectory coordinates corresponding to the work track based on the cleaning rectangular area includes: The cleaning rectangular area is mapped to obtain the cleaning two-dimensional trajectory coordinates corresponding to the working track; Obtain the calibration board image and its corresponding end TCP pose, and solve the initial hand-eye transformation matrix based on the end TCP pose; The physical center coordinates of the calibration board are calculated based on the initial hand-eye transformation matrix and the calibration board image, and the three-dimensional reconstruction consistency residual is calculated based on the physical center coordinates of the calibration board. Outlier points are iteratively eliminated based on the consistency residuals of the 3D reconstruction to obtain candidate sample points; The target hand-eye transformation matrix is ​​obtained based on the candidate sample points; The two-dimensional trajectory coordinates of the cleaning process are transformed according to the target hand-eye transformation matrix to obtain the three-dimensional trajectory coordinates of the cleaning process, which are then used as the cleaning trajectory coordinates.

[0009] In an optional implementation, determining the set of work track coordinates based on the global row spacing parameter, the center lines of all the hole rows, and the average height parameter of the hole includes: Based on the center lines of all the aforementioned hole rows, obtain the longitudinal coordinates of the exact midpoint of the center lines of all adjacent hole rows; Determine the absolute centerline coordinates based on the longitudinal coordinates of the exact midpoint of the centerlines of all adjacent holes; The safe cleaning height is determined based on the global row spacing parameter and the average hole height parameter. The set of coordinates for the work track is determined based on the absolute centerline coordinates and the safe cleaning height.

[0010] In an optional implementation, the step of calculating the global tilt angle based on the circumscribed rectangle parameters corresponding to each of the hole connected regions includes: Calculate the geometric centroid coordinates of each of the connected regions of the holes based on the circumscribed rectangle parameters of each of the connected regions of the holes. Multiple hole pairs are determined from each hole connected domain, and each hole pair satisfies a preset hole lateral spacing constraint and hole longitudinal offset constraint. The included angle is calculated based on the geometric centroid coordinates and circumscribed rectangle parameters of each hole pair to obtain the candidate included angles for each hole pair. The global tilt angle is obtained by statistically analyzing each of the candidate angles.

[0011] In a second aspect, the present invention provides an anode mud cleaning path planning device, the device comprising: The acquisition module is used to acquire the original image of the anode mud in the target area and extract multiple pore connected regions and their corresponding bounding rectangle parameters from the original image of the anode mud. The correction module is used to calculate the global tilt angle based on the circumscribed rectangle parameters corresponding to each of the hole connected regions, and to perform geometric correction on the original image of the anode mud based on the global tilt angle to obtain the corrected image. The determination module is used to obtain global row spacing parameters based on the corrected image, extract the center lines of all holes in the corrected image based on the global row spacing parameters, calculate the average height parameter of the holes based on the bounding rectangle parameters corresponding to the connected domains of each hole, and determine the set of work track coordinates based on the global row spacing parameters, the center lines of all holes, and the average height parameter of the holes. The segmentation module is used to obtain a single-channel grayscale image based on the corrected image, and to perform single-threshold segmentation based on the single-channel grayscale image to obtain a candidate anode mud binary mask; The constraint module is used to perform physical morphology constraints based on multiple connected domains of anode mud in the candidate anode mud binary mask to obtain the target anode mud binary mask; The generation module is used to generate cleaning trajectory coordinates based on the set of work trajectory coordinates and the binary mask of the target anode mud.

[0012] Thirdly, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, and the computer program, when executed by the processor, implements the anode mud cleaning path planning method as described in any of the foregoing embodiments.

[0013] Fourthly, the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the anode mud cleaning path planning method as described in any of the foregoing embodiments.

[0014] This invention discloses an anode mud cleaning path planning method, apparatus, equipment, and storage medium. The method involves acquiring an original image of anode mud in a target area, extracting multiple connected domains of holes and their corresponding bounding rectangle parameters from the original image, calculating a global tilt angle based on the bounding rectangle parameters of each connected domain, and geometrically correcting the original image of the anode mud based on the global tilt angle to obtain a corrected image. A global row spacing parameter is obtained from the corrected image, and the center lines of all hole rows in the corrected image are extracted based on the global row spacing parameter. The average height parameter of the holes is calculated based on the bounding rectangle parameters of each connected domain, and a set of work track coordinates is determined based on the global row spacing parameter, the center lines of all hole rows, and the average height parameter. A single-channel grayscale image is obtained from the corrected image, and single-threshold segmentation is performed on the single-channel grayscale image to obtain a candidate anode mud binary mask. Physical morphology constraints are applied to multiple connected domains of anode mud in the candidate anode mud binary mask to obtain a target anode mud binary mask. Finally, cleaning trajectory coordinates are generated based on the set of work track coordinates and the target anode mud binary mask. In this way, through image geometric correction, row spacing parameter-driven hole centerline optimization, and physical morphology constraint mask optimization, the robustness and accuracy of anode mud hole identification and cleaning trajectory planning are significantly improved. Global tilt angle correction eliminates shooting distortion, equidistant grid deduction compensates for incomplete hole structures, minimum operating height constraint ensures safe passage of the cleaner, and morphology constraint filters out false holes to avoid incorrect cleaning. The final generated cleaning trajectory coordinates combine spatial continuity, process safety, and equipment accessibility, greatly reducing the frequency of manual intervention and improving the automation level and operational efficiency of online cleaning of electrolytic cell anode mud. Attached Figure Description

[0015] To more clearly illustrate the technical solution of the present invention, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope of protection of the present invention. In the various drawings, similar components are numbered similarly.

[0016] Figure 1 A flowchart illustrating the anode mud cleaning path planning method proposed in this embodiment is shown. Figure 2 This embodiment shows a schematic diagram of the target area captured by the industrial camera. Figure 3 Another flowchart of the anode mud cleaning path planning method proposed in this embodiment is shown; Figure 4 A schematic diagram of the target anode mud binary mask proposed in this embodiment is shown; Figure 5 This diagram illustrates another process flow of the anode mud cleaning path planning method proposed in this embodiment. Figure 6 A schematic diagram of the cleaning rectangular area proposed in this embodiment is shown; Figure 7 A schematic diagram of another process for the anode mud cleaning path planning method proposed in this embodiment is shown; Figure 8 A schematic diagram of the anode mud cleaning path planning device proposed in this embodiment is shown.

[0017] Explanation of reference numerals in the attached diagram: 201-Zinc anode plate; 202-Industrial camera; 800-Anode mud cleaning path planning device; 801-Acquisition module; 802-Correction module; 803-Determination module; 804-Segmentation module; 805-Constraint module; 806-Generation module. Detailed Implementation

[0018] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0019] The components of the embodiments of the invention described and illustrated herein can typically be arranged and designed in various different configurations. 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.

[0020] In the following, the terms “comprising,” “having,” and their cognates, which may be used in various embodiments of the invention, are intended only to indicate a particular feature, number, step, operation, element, component, or combination thereof, and should not be construed as excluding, firstly, the presence of one or more other features, numbers, steps, operations, elements, components, or combinations thereof, or adding the possibility of one or more features, numbers, steps, operations, elements, components, or combinations thereof.

[0021] Furthermore, the terms "first," "second," and "third" are used only to distinguish descriptions and should not be interpreted as indicating or implying relative importance.

[0022] Unless otherwise specified, all terms used herein (including technical and scientific terms) shall have the same meaning as commonly understood by one of ordinary skill in the art to which the various embodiments of the invention pertain. Terms (such as those defined in commonly used dictionaries) shall be interpreted as having the same meaning as in their contextual meaning in the relevant technical field and shall not be interpreted as having an idealized or overly formal meaning, unless clearly defined in the various embodiments of the invention.

[0023] Example 1 This disclosure provides an anode mud cleaning path planning method for safely deriving a cleaning operation track with zero collision risk even when edge hole features are incomplete.

[0024] Please see Figure 1 The anode mud cleaning path planning method includes steps S101 to S106, and each step is described in detail below.

[0025] Step S101: Obtain the original image of the anode mud in the target area, and extract multiple pore connected regions and their corresponding circumscribed rectangle parameters from the original image of the anode mud.

[0026] In this embodiment, a color original image of the anode mud in the target area is acquired. After preliminary grayscale thresholding and morphological processing, multiple connected domains of holes and their corresponding bounding rectangle parameters are extracted from the original image of the anode mud.

[0027] It should be noted that the zinc anode plate to be cleaned is fixed on the suspension device, and a six-degree-of-freedom robotic arm is used as the actuator. An industrial camera is rigidly fixed to the flange at the end of the robotic arm by a clamp, thus constructing an Eye-in-Hand vision guidance configuration system so that the camera's field of view is directly facing the surface area of ​​the anode plate.

[0028] The system uses the robot arm base coordinate system as the global reference coordinate system. The spatial pose of the robot arm's end effector is represented by a six-dimensional vector, denoted as P = [X, Y, Z, Rx, Ry, Rz]. T Where X, Y, and Z represent the three-dimensional position of the end-effector center point (TCP), and Rx, Ry, and Rz represent the corresponding Euler angles and poses. Before the formal operation, the intrinsic parameters of the industrial camera were calibrated using the Zhang Zhengyou calibration method. The camera's x-axis and y-axis focal lengths (f_x, f_y) and principal point coordinates (c_x, c_y) were obtained and recorded, establishing the intrinsic parameter model for the transformation from the circumscribed rectangle parameter system to the camera coordinate system.

[0029] At the start of the operation, the robotic arm is driven by the robotic arm control box to move to the preset observation pose (i.e., the teaching point), such as... Figure 2As shown, in this pose, the industrial camera 202 acquires the original image of the anode mud on the zinc anode plate 201 in the target area and transmits it to the industrial control computer via communication links such as Ethernet. After completing subsequent algorithm processing, the industrial control computer sends the calculated three-dimensional physical coordinate sequence of the cleaning path back to the robotic arm control box, thereby guiding the robotic arm end effector to perform precise cleaning actions.

[0030] Step S102: Calculate the global tilt angle based on the circumscribed rectangle parameters corresponding to each of the hole connected regions, and perform geometric correction on the original image of the anode mud based on the global tilt angle to obtain the corrected image.

[0031] In this embodiment, the centroid of the hole is extracted based on the circumscribed rectangle parameters corresponding to each hole's connected domain to calculate the global tilt angle. The global tilt angle is then used to perform geometric correction on the original image of the anode mud to obtain the corrected image. This achieves adaptive image tilt angle correction without calibration assistance, eliminating systematic mesh distortion caused by slight deviations in the robotic arm's pose or installation skew, and laying a geometrically consistent foundation for subsequent row spacing extraction and theoretical centerline inversion.

[0032] It should be noted that the two-dimensional affine rotation transformation matrix is ​​constructed with the image center as the reference point. Geometric correction is performed on the original image of the anode mud using the origin, so that the direction of the pores is strictly aligned with the horizontal direction of the image.

[0033] In one specific embodiment, step S102 includes: calculating the geometric centroid coordinates of each of the hole connected domains based on the circumscribed rectangle parameters corresponding to each of the hole connected domains; determining multiple hole pairs from each of the hole connected domains, wherein each hole pair satisfies a preset hole lateral spacing constraint and a hole longitudinal offset constraint; calculating the included angle based on the geometric centroid coordinates and circumscribed rectangle parameters corresponding to each hole pair to obtain candidate included angles corresponding to each hole pair; and performing statistical analysis on each candidate included angle to obtain the global tilt angle.

[0034] In this embodiment, for each hole connection region, let the first... The outer rectangle parameter of the connected region of each hole is: Calculate its geometric centroid coordinates , and This represents the coordinates of the top-left corner of the circumscribed rectangle. and This represents the width and height of the bounding rectangle.

[0035] Understandably, replacing the ideal circle center with the centroid of the circumscribed rectangle of the connected domain avoids the center offset error caused by the non-closed hole and non-circular shape, thus improving the physical rationality of the centroid positioning.

[0036] Furthermore, multiple hole pairs are determined from each hole connected domain. Each hole pair is any pair of holes within the same hole row that satisfies preset lateral spacing constraints and longitudinal offset constraints. For each hole pair ( and ), calculate the angle between the line connecting the holes and the horizontal direction based on the geometric centroid coordinates and circumscribed rectangle parameters. The candidate included angles corresponding to the holes are obtained. , In the formula, ( )and( ) are the coordinates of the top-left corners of the bounding rectangles corresponding to the two holes in the hole pair, respectively. It can be understood that by introducing horizontal spacing constraints (to ensure holes in the same row) and vertical offset constraints (to eliminate mismatches across rows), reliable hole pairs with high geometric consistency are automatically selected from a large number of hole combinations, so that the tilt angle estimation is free from single-point noise interference.

[0037] Furthermore, statistical analysis was performed on each candidate angle, and the median was taken as the global tilt angle of the image. .

[0038] Understandably, using the median instead of the mean to aggregate candidate angles effectively suppresses outlier bias caused by abnormal hole pairs (such as edge-distorted holes and adhered pseudo-holes), giving the global tilt angle strong anti-interference capability and repeatability stability. It should be noted that the geometric correction, i.e., the rotation mapping formula for the circumscribed rectangle parameters, is: In the formula, The parameters of the bounding rectangle after rotation mapping. The parameters of the circumscribed rectangle before rotation mapping. Step S103: Obtain global row spacing parameters based on the corrected image; extract the center lines of all holes in the corrected image based on the global row spacing parameters; calculate the average height parameter of the holes based on the circumscribed rectangle parameters corresponding to the connected regions of each hole; and determine the coordinate set of the work track based on the global row spacing parameters, the center lines of all holes, and the average height parameter of the holes.

[0039] In this embodiment, robust global row spacing and average hole height parameters are extracted by correcting the image, and the center lines of complete hole rows and incomplete hole rows derived from the global row spacing are fused to dynamically generate a safe operation track, thereby achieving automated cleaning path planning with high coverage.

[0040] It should be noted that, based on the global line spacing parameters extracted from the corrected image, the incomplete and missing hole rows at the top and bottom edges of the image caused by the truncation of the camera's field of view are further analyzed using the center of the complete hole row in the middle of the corrected image as a reference. The incomplete hole rows at the top and bottom of the corrected image caused by the truncation of the field of view are then analyzed by using the global line spacing parameters to perform equidistant grid reverse deduction, and the center line of the missing hole row is obtained.

[0041] Accordingly, the center lines of complete hole rows are extracted from the corrected image and combined with the center lines of incomplete hole rows to form the center lines of all hole rows in the corrected image.

[0042] Understandably, the equidistant grid reverse inference elevates this parameter to prior knowledge, actively fills in the positions of invisible missing holes in the image, and generates a continuous, uniform, and physically interpretable theoretical center line. This fundamentally solves the problem of missing obstacle avoidance boundaries caused by edge blind spots, achieving full coverage of planning even if what is seen is not all.

[0043] In the corrected image, the longitudinal center position of each row of holes is extracted in the middle (i.e., the area where the holes are intact). The longitudinal distance between the centers of the kth and (k+1)th adjacent rows of holes is calculated to obtain the longitudinal distance P between the centers of adjacent rows of holes. k All valid P k The median is used as the global row spacing parameter P. global .

[0044] Simultaneously, based on the bounding rectangle parameters of all connected domains of holes, the heights of the bounding rectangles of all connected domains of holes are calculated, and the median is taken as the average height parameter H of the holes. hole .

[0045] In one specific embodiment, step S103 includes: obtaining the longitudinal coordinates of the center of all adjacent hole rows based on the center lines of all the hole rows; determining the absolute center line coordinates based on the longitudinal coordinates of the center of all adjacent hole rows; determining the safe cleaning height based on the global row spacing parameter and the average hole height parameter; and determining the set of work track coordinates based on the absolute center line coordinates and the safe cleaning height.

[0046] In this embodiment, adjacent hole rows are two adjacent rows of holes. For each pair of adjacent rows of holes, the arithmetic mean of their longitudinal coordinates is taken at the exact midpoint of the center lines of the two hole rows (including the center line of the complete hole row and the deduced center line of the incomplete hole row) as the absolute center line coordinate of the working track, ensuring that all working tracks are strictly limited to the safe area between the two adjacent rows of holes.

[0047] Meanwhile, in order to completely avoid the engineering risk of robotic cleaning tools colliding with the hole wall during rigid motion from a mathematical perspective, the global row spacing parameter P...global An obstacle avoidance and clearance deduction mechanism is constructed across the spatial span to establish the safe cleaning height H corresponding to each working track. r Its dynamic solution rule satisfies: Hr=P global -H hole -2ɑ, where ɑ is the preset single-sided electromechanical anti-collision safety margin pixel value. This step calculates the limit safe cleaning width of the pure metal blank area between the two rows of holes by continuously subtracting the height of the hole itself and the physical buffer distance. Then, it outputs the work track set {y} by combining the absolute centerline coordinates and the safe cleaning height. l}

[0048] It should be noted that during the extrapolation generation of the work track, if the theoretical calculation boundary of the edge blind zone exceeds the actual physical field of view of the image (i.e., the top upper boundary y < 0 or the bottom lower boundary y > H), it is very easy for the robotic arm control system to report an error due to pixel index out-of-bounds. To address this, a track adaptive clipping and centerline reconstruction mechanism based on the field of view boundary is introduced: when the algorithm detects that the edge boundary of the theoretical track overflows, the system immediately performs hard truncation of the boundary, forcibly pulling the boundary at the out-of-bounds point back and fitting it to the outermost edge of the image (i.e., the top out-of-bounds boundary y > H). top = 0, bottom out of bounds, let y bottom = H, where H is the total height of the image), while keeping the theoretical boundary coordinates of the side of the track that has not crossed the boundary completely unchanged. At this time, because the overflow area is forcibly retracted, the working track completes adaptive shrinkage and clipping at the physical boundary. The system then takes the geometric center coordinates of the actual upper and lower boundaries after clipping, and recalculates and draws the center reference line of the track at the new intermediate position. Through this dynamic correction mechanism, it is ensured that no matter how the electrode workpiece is truncated at the edge of the camera's field of view, the derived edge working track can achieve a safe limit closed loop within the physical field of view, which not only completely eliminates the risk of the control system crossing the boundary, but also maximizes the safe coverage of residual mud spots at the edge.

[0049] Step S104: Obtain a single-channel grayscale image based on the corrected image, and perform single-threshold segmentation based on the single-channel grayscale image to obtain a candidate anode mud binary mask.

[0050] In this embodiment, the corrected color image is converted to the HSV color space, and the luminance channel V is extracted as the target for enhancement processing.

[0051] Furthermore, Limit Contrast Adaptive Histogram Equalization (CLAHE) is employed to suppress specular artifacts and shadows caused by point light sources, resulting in an enhanced brightness image. Then set the grayscale threshold. A preliminary binary mask for anode mud is generated through single-threshold segmentation. : .

[0052] Furthermore, to eliminate small-area fractures caused by local noise, a k×k convolution kernel K is used to perform morphological closing operations on the preliminary binary mask, filling small holes and smoothing region edges. The candidate anode mud binary mask after morphological closing is then obtained. The expression is: ,in, This indicates an expansion operation. This indicates a corrosion operation.

[0053] Understandably, performing segmentation on the geometrically corrected image eliminates the effect of tilt distortion on the uniformity of the threshold response, making holes in the same row more consistent in grayscale response.

[0054] Step S105: Physical morphology constraints are applied to multiple connected domains of anode mud in the candidate anode mud binary mask to obtain the target anode mud binary mask.

[0055] In this embodiment, physical morphology constraints are applied to each connected domain of the candidate anode mud in the binary mask to obtain a high-confidence target anode mud binary mask M. final Physical shape constraints include area constraints, roughness constraints, and contour complexity constraints.

[0056] Please see Figure 3 In one specific embodiment, step S105 includes steps S1051 to S1053, and each step is described in detail below.

[0057] Step S1051: Calculate the pixel area of ​​each connected region of the anode mud in the candidate anode mud binary mask, and perform anode mud connected region filtering according to the preset area threshold and the pixel area of ​​each connected region of the anode mud to obtain the first filtered binary mask.

[0058] In this embodiment, the pixel area of ​​each connected region of anode mud in the candidate anode mud binary mask is calculated. A i , pixel area A i Less than the preset area threshold The connected components are discarded as discrete debris to obtain the first-screened binary mask, which retains the pixel area range corresponding to the standard aperture of the zinc anode plate, thereby improving the true positive rate of hole recognition.

[0059] Step S1052: Calculate the local variance of gray values ​​of each connected domain of anode mud in the first filtered binary mask, and perform anode mud connected domain screening according to the preset roughness threshold and the local variance of gray values ​​of each connected domain of anode mud to obtain the second filtered binary mask.

[0060] In this embodiment, the local variance of the grayscale values ​​of each anode mud connected domain in the binary mask after the first screening is calculated. Because the surface roughness of real anode mud blocks leads to a large variance, if the local variance of grayscale values ​​is high... Less than or equal to the preset roughness threshold If the image is smooth water stain or a reflection artifact on the base plate, it is identified as such and removed. This results in a second-filtered binary mask, which filters out false holes caused by blurred edges, partial occlusion, and uneven lighting through local variance filtering, thereby enhancing the credibility of the mask's spatial structure.

[0061] Step S1053: Calculate the density ratio of the number of edge pixels of each anode mud connected region to the total pixel area inside the connected region in the second filtered binary mask. Based on the preset complexity threshold and the density ratio of each anode mud connected region, filter the anode mud connected regions to obtain the target anode mud binary mask.

[0062] In this embodiment, the density ratio of the number of pixels at the edge of each anode mud connected region in the second-screened binary mask to the total pixel area inside the connected region is calculated. If the density ratio Less than the preset complexity threshold If the oxide scale is elongated or blurred, it is identified as interference and removed to obtain the target binary anode mud mask. M final ,like Figure 4 As shown, this effectively distinguishes between real circular holes (low ratio) and elongated cracks and strip-shaped stains (high ratio), achieving essential filtering at the morphological level.

[0063] Step S106: Generate cleaning trajectory coordinates based on the set of work trajectory coordinates and the binary mask of the target anode mud.

[0064] In this embodiment, the effective cleaning trajectory coordinates are determined from the set of operation trajectory coordinates by utilizing the high-confidence information of the effective hole geometry and structural integrity of the anode plate characterized by the binary mask of the target anode mud. In the case of incomplete edge hole features, a cleaning operation trajectory with zero collision risk is safely deduced.

[0065] Please see Figure 5 In one specific embodiment, step S106 includes steps S1061 to S1064, and each step is described in detail below.

[0066] Step S1061: For each work track in the work track coordinate set, extract a strip area with a preset longitudinal height along the work track.

[0067] In this embodiment, for each work track in the work track coordinate set A preset longitudinal height is cut along the working track. The strip area has a preset vertical height, which is the safe cleaning height.

[0068] Step S1062: Determine the mask region corresponding to the strip region based on the binary mask of the target anode mud, and perform column direction integration on the mask region to obtain the pixel density distribution.

[0069] In this embodiment, the pixel density distribution function D(x) is expressed as: In the formula, The pixel value (0 or 1) of the target anode mud binary mask at coordinates (x, y).

[0070] Understandably, compressing the two-dimensional mask information of the anode mud into a one-dimensional pixel density distribution significantly reduces the computational dimension while preserving the spatial distribution characteristics of the holes along the track direction.

[0071] Step S1063: Traverse the pixel density distribution according to the preset start and end density thresholds to obtain the target segment corresponding to the operation track.

[0072] In this embodiment, the pixel density in the pixel density distribution is traversed and compared according to the preset start and end density thresholds to select the target section corresponding to the working track. This effectively addresses the density fluctuations caused by local accumulation of anode mud, avoids trajectory jumps caused by single-point noise, and improves the robustness of section positioning.

[0073] In one specific embodiment, the preset start and end density thresholds include a preset start density threshold and a preset end density threshold. Step S1063 includes: when traversing the pixel density distribution until the current pixel density is greater than the preset start density threshold, then the pixel corresponding to the current pixel density is taken as the start point of the target segment; when traversing the pixel density distribution until the current pixel density is less than the preset end density threshold, and the state where the current pixel density is less than the preset end density threshold exceeds a preset distance, then the pixel corresponding to the current pixel density distribution is taken as the end point of the target segment.

[0074] In this embodiment, the pixel density distribution is traversed along the positive x-axis until the current pixel density is greater than a preset starting density threshold. At that time, the coordinates corresponding to the current pixel density are marked as the starting point of the target segment. .

[0075] Furthermore, in recording Then continue traversing the pixel density distribution until the current pixel density is less than the preset termination density threshold. And the current pixel density is less than the preset termination density threshold. The state exceeds the preset distance W drop When the pixel count reaches a certain value, a truncation is triggered, and the current coordinates are moved back by W. drop The end point of the target segment is marked later. .

[0076] Step S1064: Based on the preset rectangular horizontal expansion compensation pixels and the preset vertical height, the target segment is restored to a two-dimensional rectangular bounding box as a cleaning rectangular area, and the cleaning trajectory coordinates corresponding to the working track are obtained by mapping the cleaning rectangular area.

[0077] In this embodiment, the target segment (x) is obtained. s , x e After that, let the horizontal expansion compensation pixel be E. Using the preset horizontal expansion compensation pixel and preset vertical height, the one-dimensional target segment is restored to a two-dimensional rectangular bounding box as the cleaning rectangular region, such as... Figure 6 As shown, its coordinates are represented as: [x s -E, y l -h r / 2, x e +E, y l +h r / 2], which is then mapped to the cleaning trajectory coordinates corresponding to the work track.

[0078] In one specific embodiment, step S1064 includes: mapping the cleaning rectangular area to obtain the cleaning two-dimensional trajectory coordinates corresponding to the working track; acquiring a calibration board image and its corresponding end-effector TCP pose, and solving the initial hand-eye transformation matrix based on the end-effector TCP pose; calculating the physical center coordinates of the calibration board based on the initial hand-eye transformation matrix and the calibration board image, and calculating the three-dimensional reconstruction consistency residual based on the physical center coordinates of the calibration board; performing outlier point iterative removal based on the three-dimensional reconstruction consistency residual to obtain candidate sample points; obtaining the target hand-eye transformation matrix based on the candidate sample points; and transforming the cleaning two-dimensional trajectory coordinates based on the target hand-eye transformation matrix to obtain the cleaning three-dimensional trajectory coordinates as the cleaning trajectory coordinates.

[0079] In this embodiment, the cleaning rectangular area is first mapped back to the original image coordinate system, consistent with the rotation correction operation (inverse affine transformation), to obtain the cleaning two-dimensional trajectory coordinates. The final pixel trajectory coordinate array is then transmitted to the industrial control computer. At this point, the generated path instructions still belong to the pixel space of the two-dimensional image and cannot be directly executed by the robot arm's underlying controller.

[0080] Furthermore, because the anode plate is vertically suspended, the movement of the robotic arm is restricted within the narrow working space (limited by large tilt angles and rotation angles), leading to degradation errors in traditional calibration. To address this, the robotic arm is controlled to move to multiple poses within this confined space, images of the calibration plate are acquired, and the end-effector TCP pose is recorded simultaneously. A set of hand-eye calibration equations based on AX=XB is constructed, and the initial hand-eye transformation matrix is ​​initially solved using the Tsai algorithm. .

[0081] Furthermore, based on the initially obtained initial hand-eye transformation matrix... The spatial coordinates of the calibration plate calculated by vision under each posture are uniformly mapped to the coordinate system of the robot arm base to obtain the physical center coordinates of the calibration plate under the i-th posture. Calculate all ( The physical spatial geometric center of the sample The three-dimensional Euclidean distance of each sample point relative to the geometric center of all samples is used to calculate the three-dimensional reconstruction consistency residual. .

[0082] Furthermore, the top-level samples with the largest 3D reconstruction consistency residuals (such as the top 10% of abnormal pose data) are successively removed, and the calibration matrix is ​​resolved using the remaining high spatial consistency (high contribution) samples. This process is continued iteratively until the overall root mean square error (RMSE) converges and stabilizes. Figure 7 As shown, a high-precision target hand-eye transformation matrix is ​​obtained, which avoids the degradation problem in confined space, providing an accurate mathematical basis for the final execution of robotic arm movements.

[0083] Furthermore, based on the target hand-eye transformation matrix and the fixed working depth parameters of the camera, the two-dimensional cleaning trajectory coordinates are converted into three-dimensional cleaning trajectory coordinates in the coordinate system of the robotic arm base. These coordinates serve as the final cleaning trajectory coordinates. The flexible blade at the end of the robotic arm is then controlled to perform continuous scraping along the converted three-dimensional physical path. This integrates the anti-collision holes and zero-empty stroke design at the algorithm level into physical actions, completing the closed loop of the entire system's operation.

[0084] The anode mud cleaning path planning method proposed in this embodiment acquires an original image of the anode mud in the target area, extracts multiple connected domains of holes and their corresponding bounding rectangle parameters from the original anode mud image; calculates a global tilt angle based on the bounding rectangle parameters corresponding to each connected domain of holes, and performs geometric correction on the original anode mud image based on the global tilt angle to obtain a corrected image; obtains a global row spacing parameter based on the corrected image, extracts the center lines of all hole rows in the corrected image based on the global row spacing parameter, calculates the average height parameter of holes based on the bounding rectangle parameters corresponding to each connected domain of holes, and determines the work track coordinate set based on the global row spacing parameter, the center lines of all hole rows, and the average height parameter of holes; obtains a single-channel grayscale image based on the corrected image, performs single-threshold segmentation based on the single-channel grayscale image to obtain a candidate anode mud binary mask; performs physical morphology constraints on multiple connected domains of anode mud in the candidate anode mud binary mask to obtain a target anode mud binary mask; and generates cleaning trajectory coordinates based on the work track coordinate set and the target anode mud binary mask. In this way, through image geometric correction, row spacing parameter-driven hole centerline optimization, and physical morphology constraint mask optimization, the robustness and accuracy of anode mud hole identification and cleaning trajectory planning are significantly improved. Global tilt angle correction eliminates shooting distortion, equidistant grid deduction compensates for incomplete hole structures, minimum operating height constraint ensures safe passage of the cleaner, and morphology constraint filters out false holes to avoid incorrect cleaning. The final generated cleaning trajectory coordinates combine spatial continuity, process safety, and equipment accessibility, greatly reducing the frequency of manual intervention and improving the automation level and operational efficiency of online cleaning of electrolytic cell anode mud.

[0085] Example 2 Furthermore, this disclosure provides an anode mud cleaning path planning device 800, please refer to [link to relevant documentation]. Figure 8 The device includes: The acquisition module 801 is used to acquire the original image of the anode mud in the target area and extract multiple pore connected regions and their corresponding bounding rectangle parameters from the original image of the anode mud. The correction module 802 is used to calculate the global tilt angle based on the circumscribed rectangle parameters corresponding to each of the hole connected regions, and to perform geometric correction on the original image of the anode mud based on the global tilt angle to obtain the corrected image. The determination module 803 is used to obtain global row spacing parameters based on the corrected image, extract the center lines of all holes in the corrected image based on the global row spacing parameters, calculate the average height parameter of the holes based on the bounding rectangle parameters corresponding to each hole's connected domain, and determine the set of work track coordinates based on the global row spacing parameters, all hole row center lines, and the average height parameter of the holes. The segmentation module 804 is used to obtain a single-channel grayscale image based on the corrected image, and to perform single-threshold segmentation based on the single-channel grayscale image to obtain a candidate anode mud binary mask. The constraint module 805 is used to perform physical morphology constraints based on multiple connected domains of anode mud in the candidate anode mud binary mask to obtain the target anode mud binary mask; The generation module 806 is used to generate cleaning trajectory coordinates based on the set of work trajectory coordinates and the binary mask of the target anode mud.

[0086] Optionally, the constraint module 805 is used to calculate the pixel area of ​​each of the anode mud connected regions in the candidate binary mask, and to filter the anode mud connected regions according to a preset area threshold and the pixel area of ​​each of the anode mud connected regions to obtain a first filtered binary mask; calculate the local variance of the gray value of each anode mud connected region in the first filtered binary mask, and to filter the anode mud connected regions according to a preset roughness threshold and the local variance of the gray value of each of the anode mud connected regions to obtain a second filtered binary mask; calculate the density ratio of the number of pixels at the edge of each anode mud connected region to the total pixel area inside the connected region in the second filtered binary mask, and to filter the anode mud connected regions according to a preset complexity threshold and the density ratio of each of the anode mud connected regions to obtain the target anode mud binary mask.

[0087] Optionally, the generation module 806 is used to, for each working track in the set of working track coordinates, extract a strip region with a preset longitudinal height along the working track; determine the mask region corresponding to the strip region according to the binary mask of the target anode mud, perform column direction integration on the mask region to obtain the pixel density distribution; traverse the pixel density distribution according to a preset start and end density threshold to obtain the target segment corresponding to the working track; restore the target segment to a two-dimensional rectangular bounding box as a cleaning rectangular region according to a preset rectangular horizontal expansion compensation pixel and the preset longitudinal height, and obtain the cleaning trajectory coordinates corresponding to the working track according to the cleaning rectangular region.

[0088] Optionally, the generation module 806 is configured to, when traversing the pixel density distribution until the current pixel density is greater than the preset starting density threshold, take the pixel corresponding to the current pixel density as the starting point of the target segment; when traversing the pixel density distribution until the current pixel density is less than the preset ending density threshold, and the state where the current pixel density is less than the preset ending density threshold exceeds a preset distance, take the pixel corresponding to the current pixel density distribution as the ending point of the target segment.

[0089] Optionally, the generation module 806 is used to map the cleaning rectangular area to obtain the cleaning two-dimensional trajectory coordinates corresponding to the working track; acquire the calibration board image and its corresponding end-effector TCP pose, and solve the initial hand-eye transformation matrix based on the end-effector TCP pose; calculate the physical center coordinates of the calibration board based on the initial hand-eye transformation matrix and the calibration board image, and calculate the three-dimensional reconstruction consistency residual based on the physical center coordinates of the calibration board; perform outlier point iterative removal based on the three-dimensional reconstruction consistency residual to obtain candidate sample points; obtain the target hand-eye transformation matrix based on the candidate sample points; and transform the cleaning two-dimensional trajectory coordinates based on the target hand-eye transformation matrix to obtain the cleaning three-dimensional trajectory coordinates as the cleaning trajectory coordinates.

[0090] Optionally, the determining module 803 is used to obtain the longitudinal coordinates of the center of all adjacent hole rows based on the center lines of all the hole rows; determine the absolute center line coordinates based on the longitudinal coordinates of the center of all adjacent hole rows; determine the safe cleaning height based on the global row spacing parameter and the average hole height parameter; and determine the set of work track coordinates based on the absolute center line coordinates and the safe cleaning height.

[0091] Optionally, the correction module 802 is used to calculate the geometric centroid coordinates of each of the hole connected regions based on the circumscribed rectangle parameters corresponding to each of the hole connected regions; determine multiple hole pairs from each of the hole connected regions, each hole pair satisfying a preset hole lateral spacing constraint and hole longitudinal offset constraint; calculate the included angle based on the geometric centroid coordinates and circumscribed rectangle parameters corresponding to each hole pair to obtain the candidate included angles corresponding to each hole pair; and perform statistical analysis on each of the candidate included angles to obtain the global tilt angle.

[0092] The apparatus provided in this embodiment can execute the steps of the anode mud cleaning path planning method provided in Embodiment 1. To avoid repetition, it will not be described again.

[0093] The anode mud cleaning path planning device proposed in this embodiment acquires an original image of anode mud in a target area, extracts multiple connected domains of holes and their corresponding bounding rectangle parameters from the original image, calculates a global tilt angle based on the bounding rectangle parameters corresponding to each connected domain of holes, performs geometric correction on the original image of anode mud based on the global tilt angle, and obtains a corrected image. It then acquires a global row spacing parameter based on the corrected image, extracts the center lines of all hole rows in the corrected image based on the global row spacing parameter, calculates the average height parameter of the holes based on the bounding rectangle parameters corresponding to each connected domain of holes, and determines a set of work track coordinates based on the global row spacing parameter, the center lines of all hole rows, and the average height parameter of the holes. Finally, it obtains a single-channel grayscale image based on the corrected image, performs single-threshold segmentation based on the single-channel grayscale image, and obtains a candidate anode mud binary mask. It then performs physical morphology constraints on multiple connected domains of anode mud in the candidate anode mud binary mask to obtain a target anode mud binary mask. Finally, it generates cleaning trajectory coordinates based on the set of work track coordinates and the target anode mud binary mask. In this way, through image geometric correction, row spacing parameter-driven hole centerline optimization, and physical morphology constraint mask optimization, the robustness and accuracy of anode mud hole identification and cleaning trajectory planning are significantly improved. Global tilt angle correction eliminates shooting distortion, equidistant grid deduction compensates for incomplete hole structures, minimum operating height constraint ensures safe passage of the cleaner, and morphology constraint filters out false holes to avoid incorrect cleaning. The final generated cleaning trajectory coordinates combine spatial continuity, process safety, and equipment accessibility, greatly reducing the frequency of manual intervention and improving the automation level and operational efficiency of online cleaning of electrolytic cell anode mud.

[0094] Example 3 Furthermore, this disclosure provides a computer device including a memory and a processor. The memory stores a computer program, which, when executed by the processor, implements the anode mud cleaning path planning method described in Embodiment 1.

[0095] The device provided in this embodiment can perform the steps of the anode mud cleaning path planning method provided in Embodiment 1. To avoid repetition, it will not be described again.

[0096] Example 4 This disclosure provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the anode mud cleaning path planning method described in Embodiment 1.

[0097] In this embodiment, the computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, etc.

[0098] The computer-readable storage medium provided in this embodiment can implement the anode mud cleaning path planning method provided in Embodiment 1. To avoid repetition, it will not be described again here.

[0099] In all examples shown and described herein, any specific values ​​should be interpreted as merely exemplary and not as limitations; therefore, other examples of exemplary embodiments may have different values.

[0100] 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.

[0101] The above-described embodiments are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention.

Claims

1. A method for planning anode mud cleaning path, characterized in that, The method includes: Acquire the original image of the anode mud in the target area, and extract multiple connected domains of pores and their corresponding bounding rectangle parameters from the original image of the anode mud. The global tilt angle is calculated based on the circumscribed rectangle parameters corresponding to each of the hole connected regions. The original image of the anode mud is then geometrically corrected based on the global tilt angle to obtain the corrected image. The global row spacing parameter is obtained based on the corrected image. The center lines of all holes in the corrected image are extracted based on the global row spacing parameter. The average height parameter of the hole is calculated based on the bounding rectangle parameter corresponding to each hole's connected region. The coordinate set of the work track is determined based on the global row spacing parameter, the center lines of all holes, and the average height parameter of the hole. A single-channel grayscale image is obtained based on the corrected image, and single-threshold segmentation is performed based on the single-channel grayscale image to obtain a candidate anode mud binary mask; The target anode mud binary mask is obtained by physically constraining multiple connected domains of anode mud in the candidate anode mud binary mask; The cleaning trajectory coordinates are generated based on the set of work trajectory coordinates and the binary mask of the target anode mud.

2. The anode mud cleaning path planning method according to claim 1, characterized in that, The step of determining the work track coordinate set based on the global row spacing parameter, the center lines of all the holes, and the average height parameter of the holes includes: Based on the center lines of all the aforementioned hole rows, obtain the longitudinal coordinates of the exact midpoint of the center lines of all adjacent hole rows; Determine the absolute centerline coordinates based on the longitudinal coordinates of the exact midpoint of the centerlines of all adjacent holes; The safe cleaning height is determined based on the global row spacing parameter and the average hole height parameter. The set of coordinates for the work track is determined based on the absolute centerline coordinates and the safe cleaning height.

3. The anode mud cleaning path planning method according to claim 1, characterized in that, The step of obtaining the target anode mud binary mask by physically constraining multiple connected domains of anode mud in the candidate anode mud binary mask includes: Calculate the pixel area of ​​each connected region of anode mud in the candidate anode mud binary mask, and filter the connected regions of anode mud according to the preset area threshold and the pixel area of ​​each connected region of anode mud to obtain the first filtered binary mask. Calculate the local variance of gray values ​​of each anode mud connected domain in the first filtered binary mask, and perform anode mud connected domain filtering according to the preset roughness threshold and the local variance of gray values ​​of each anode mud connected domain to obtain the second filtered binary mask. Calculate the density ratio of the number of pixels at the edge of each connected domain of the anode mud in the second filtered binary mask to the density ratio of the total pixel area inside the connected domain. Based on the preset complexity threshold and the density ratio of each connected domain of the anode mud, filter the connected domains of the anode mud to obtain the target anode mud binary mask.

4. The anode mud cleaning path planning method according to claim 1, characterized in that, The step of generating cleaning trajectory coordinates based on the set of work trajectory coordinates and the binary mask of the target anode mud includes: For each work track in the set of work track coordinates, a strip region with a preset longitudinal height is extracted along the work track; The mask region corresponding to the strip region is determined based on the binary mask of the target anode mud, and the pixel density distribution is obtained by performing column direction integration on the mask region. The pixel density distribution is traversed according to a preset start and end density threshold to obtain the target segment corresponding to the work track. The target segment is restored into a two-dimensional rectangular bounding box as the cleaning rectangular area based on the preset rectangular horizontal expansion compensation pixels and the preset vertical height. The cleaning trajectory coordinates corresponding to the operation track are then obtained by mapping the cleaning rectangular area.

5. The anode mud cleaning path planning method according to claim 4, characterized in that, The preset start and end density thresholds include a preset start density threshold and a preset end density threshold. The step of traversing the pixel density distribution according to the preset start and end density thresholds to obtain the target segment corresponding to the work track includes: When the pixel density distribution is traversed until the current pixel density is greater than the preset starting density threshold, the pixel corresponding to the current pixel density is taken as the starting point of the target segment. When the pixel density distribution is traversed until the current pixel density is less than the preset termination density threshold, and the state where the current pixel density is less than the preset termination density threshold exceeds a preset distance, then the pixel point corresponding to the current pixel density distribution is taken as the endpoint of the target segment.

6. The anode mud cleaning path planning method according to claim 4, characterized in that, The step of obtaining the cleaning trajectory coordinates corresponding to the work track based on the mapping of the cleaning rectangular area includes: The cleaning rectangular area is mapped to obtain the cleaning two-dimensional trajectory coordinates corresponding to the working track; Obtain the calibration board image and its corresponding end TCP pose, and solve the initial hand-eye transformation matrix based on the end TCP pose; The physical center coordinates of the calibration board are calculated based on the initial hand-eye transformation matrix and the calibration board image, and the three-dimensional reconstruction consistency residual is calculated based on the physical center coordinates of the calibration board. Outlier points are iteratively eliminated based on the consistency residuals of the 3D reconstruction to obtain candidate sample points; The target hand-eye transformation matrix is ​​obtained based on the candidate sample points; The two-dimensional trajectory coordinates of the cleaning process are transformed according to the target hand-eye transformation matrix to obtain the three-dimensional trajectory coordinates of the cleaning process, which are then used as the cleaning trajectory coordinates.

7. The anode mud cleaning path planning method according to claim 1, characterized in that, The calculation of the global tilt angle based on the circumscribed rectangle parameters corresponding to each of the hole connected regions includes: Calculate the geometric centroid coordinates of each of the connected regions of the holes based on the circumscribed rectangle parameters of each of the connected regions of the holes. Multiple hole pairs are determined from each hole connected domain, and each hole pair satisfies a preset hole lateral spacing constraint and hole longitudinal offset constraint. The included angle is calculated based on the geometric centroid coordinates and circumscribed rectangle parameters of each hole pair to obtain the candidate included angles for each hole pair. The global tilt angle is obtained by statistically analyzing each of the candidate angles.

8. A path planning device for cleaning anode mud, characterized in that, The device includes: The acquisition module is used to acquire the original image of the anode mud in the target area and extract multiple pore connected regions and their corresponding bounding rectangle parameters from the original image of the anode mud. The correction module is used to calculate the global tilt angle based on the circumscribed rectangle parameters corresponding to each of the hole connected regions, and to perform geometric correction on the original image of the anode mud based on the global tilt angle to obtain the corrected image. The determination module is used to obtain global row spacing parameters based on the corrected image, extract the center lines of all holes in the corrected image based on the global row spacing parameters, calculate the average height parameter of the holes based on the bounding rectangle parameters corresponding to the connected domains of each hole, and determine the set of work track coordinates based on the global row spacing parameters, the center lines of all holes, and the average height parameter of the holes. The segmentation module is used to obtain a single-channel grayscale image based on the corrected image, and to perform single-threshold segmentation based on the single-channel grayscale image to obtain a candidate anode mud binary mask; The constraint module is used to perform physical morphology constraints based on multiple connected domains of anode mud in the candidate anode mud binary mask to obtain the target anode mud binary mask; The generation module is used to generate cleaning trajectory coordinates based on the set of work trajectory coordinates and the binary mask of the target anode mud.

9. A computer device, characterized in that, It includes a memory and a processor, the memory storing a computer program that, when executed by the processor, implements the anode mud cleaning path planning method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, It stores a computer program that, when executed by a processor, implements the anode mud cleaning path planning method as described in any one of claims 1 to 7.

Images