Method, system, medium and product for track planning of laser rust removal of closing seam

Abstract

The application discloses a kind of closing joint laser rust removal operation track planning method, system, medium and product, it is related to laser rust removal technical field, the original image of closing joint area is converted from RGB color space to HSV color space in the application, and according to preset threshold, the accurate division of paint, ablation and bare metal area is completed, the accuracy and robustness of area classification under complex operation environment are improved;Meanwhile, the edge gradient of three channels is calculated respectively, the local gradient main direction is extracted, and the robust fitting and screening of boundary point are realized by combining random sampling consistent algorithm, can eliminate noise, pseudo-edge and other interference factors, obtain continuous reliable area boundary, significantly improve the defect that single visual feature boundary positioning precision is insufficient;Finally, based on accurate boundary planning laser rust removal operation track, the complete coverage and accurate rust removal of ablation area can be realized, the track control precision, operation stability and rust removal quality of ship closing joint laser rust removal operation are greatly improved.

Description

Technical Field

[0001] This application relates to the field of laser rust removal technology, and in particular to a method, system, medium and product for planning the trajectory of laser rust removal operation at closure seams. Background Technology

[0002] In shipbuilding and maintenance, rust removal at seams is a crucial step in ensuring the adhesion of subsequent paint coatings. Laser rust removal technology holds promise as a replacement for traditional manual grinding and roughening processes, but it requires extremely high precision in trajectory control: if the trajectory deviates from the target area, it may result in paint residue or incomplete rust removal; if the laser beam is over-irradiated, it may damage the substrate metal. Existing methods struggle to accurately identify area boundaries in complex seam scenarios (containing multiple areas such as painted, ablated, and bare metal, accompanied by rust, oil stains, and uneven lighting), leading to insufficient trajectory planning accuracy.

[0003] In the context of ship closure joints, existing technologies have the following drawbacks: The trajectory planning method based on grayscale images and edge detection uses an industrial camera to acquire grayscale images of the closure seam area, and combines edge detection operators such as Canny and Sobel to extract the region boundaries, thereby planning the laser rust removal trajectory. However, grayscale images have limitations in distinguishing different types of regions: due to the presence of various regions such as original paint, ablation discoloration, and bare metal, often accompanied by interference from rust and oil stains, grayscale information cannot effectively differentiate region types, resulting in low boundary recognition accuracy.

[0004] Region segmentation methods based on the RGB color space utilize color information from the three RGB channels for region segmentation. Thresholding or clustering algorithms are used to distinguish painted areas, corroded areas, and bare metal areas, and the operation trajectory is planned based on the segmentation results. However, the RGB color space is susceptible to illumination variations: the three RGB channels are highly coupled, resulting in poor robustness under different lighting conditions, making it difficult to meet the stability requirements of actual working environments.

[0005] Region recognition methods based on single visual features use a single feature (such as grayscale, color, texture, etc.) combined with a classifier for region recognition. Boundary points are obtained through edge extraction, and a laser scanning path is planned. However, there are shortcomings in boundary positioning accuracy: the boundary shape of the closure seam region is complex, with interference from noise points, false edges, etc. Traditional edge detection methods struggle to accurately extract continuous and reliable boundary point sets, affecting the accuracy of trajectory planning. Summary of the Invention

[0006] This application provides a method, system, medium, and product for laser rust removal operation trajectory planning for closure seams, which at least solves the problems of existing laser rust removal trajectory planning methods, such as the difficulty in distinguishing multiple types of regions from grayscale images, the susceptibility of RGB color space to illumination, and the insufficient accuracy of single visual feature boundary positioning, which in turn leads to inaccurate boundary recognition of closure seam areas, low trajectory planning accuracy, and the tendency for incomplete rust removal or damage to the substrate.

[0007] This application provides a method for planning the trajectory of laser rust removal operations at closure seams, including: Obtain the original image of the seam area and convert the original image from the RGB color space to the HSV color space; In the HSV color space, the seam area is divided into a painted area, an ablation area, and a bare metal area according to preset hue threshold, saturation threshold, and brightness threshold. The edge gradients of the hue channel image, saturation channel image, and brightness channel image in the horizontal and vertical directions are calculated respectively to obtain three sets of independent gradient magnitude and gradient direction information, and hue channel gradient map, saturation channel gradient map, and brightness channel gradient map are generated respectively. Based on the directional gradient histogram operator, the gradient direction histogram in the local neighborhood is extracted for each pixel in the hue channel gradient map, the saturation channel gradient map, and the brightness channel gradient map, respectively, and the peak direction of the gradient direction histogram is taken as the local gradient principal direction of the pixel. A random sampling consensus algorithm is used to iteratively fit the pixels in the hue channel gradient map, the saturation channel gradient map, and the brightness channel gradient map to form a boundary model. With the main local gradient direction as a constraint, pixels whose angle between the main local gradient direction and the normal direction of the boundary model is less than a preset angle are determined as interior points, and the interior points are used as the confirmed boundary points. Based on the confirmed boundary points, plan the working trajectory of the laser rust removal head in the closure joint area.

[0008] Furthermore, in the HSV color space, the seam area is divided into a painted area, an ablation area, and a bare metal area according to preset hue thresholds, saturation thresholds, and lightness thresholds, including: A first hue threshold is set based on the difference in hue between the paint layer and the metal substrate, and pixels whose hue values ​​fall within the range of the first hue threshold are divided into paint areas. A saturation threshold is set based on the difference in saturation between bare metal and paint layer, and pixels with saturation values ​​lower than the saturation threshold are classified as bare metal regions. A brightness threshold is set based on the difference in brightness between the ablation and discoloration areas and the original painted areas. Pixels with brightness values ​​lower than the brightness threshold and not belonging to the painted areas are classified as ablation areas.

[0009] Furthermore, the edge gradients of the hue channel image, saturation channel image, and brightness channel image are calculated in the horizontal and vertical directions respectively, resulting in three sets of independent gradient magnitude and gradient direction information, including: The Sobel operator is used to calculate the gradient magnitudes of the hue channel image, saturation channel image, and brightness channel image in the horizontal and vertical directions, respectively. Let the grayscale value of the pixel at position (x,y) in any of the three channels of the image be I_H(x,y), where x is the horizontal column coordinate of the image pixel and y is the vertical row coordinate of the image pixel; The formulas for calculating the horizontal gradient G_x at pixel (x,y) and the vertical gradient G_y at pixel (x,y) are as follows: G_x = I_H(x+1,y-1) + 2I_H(x+1,y) + I_H(x+1,y+1) - I_H(x-1,y-1) - 2I_H(x-1,y) - I_H(x-1,y+1); where I_H(x+1,y-1) is the pixel grayscale value at the position (x+1,y-1), I_H(x+1,y) is the pixel grayscale value at the position (x+1,y), I_H(x+1,y+1) is the pixel grayscale value at the position (x+1,y+1), I_H(x-1,y-1) is the pixel grayscale value at the position (x-1,y-1), I_H(x-1,y) is the pixel grayscale value at the position (x-1,y), I_H(x-1,y+1) is the pixel grayscale value at the position (x-1,y+1); G_y = I_H(x-1,y+1) + 2I_H(x,y+1) + I_H(x+1,y+1) - I_H(x-1,y-1) - 2I_H(x,y-1) - I_H(x+1,y-1); where I_H(x-1,y+1) is the pixel grayscale value corresponding to the pixel position (x-1,y+1), I_H(x,y+1) is the pixel grayscale value corresponding to the pixel position (x,y+1), I_H(x+1,y+1) is the pixel grayscale value corresponding to the pixel position (x+1,y+1), I_H(x-1,y-1) is the pixel grayscale value corresponding to the pixel position (x-1,y-1), I_H(x,y-1) is the pixel grayscale value corresponding to the pixel position (x,y-1), I_H(x+1,y-1) is the pixel grayscale value corresponding to the pixel position (x+1,y-1); The formulas for calculating the gradient magnitude M(x,y) and gradient direction θ(x,y) are as follows: M(x,y) = ; θ(x,y) = ; The gradient magnitude matrix and gradient direction matrix corresponding to the hue channel image, the saturation channel image, and the brightness channel image are obtained.

[0010] Furthermore, the gradient direction histogram operator, based on the gradient direction histogram, extracts the gradient direction histogram within the local neighborhood for each pixel in the hue channel gradient map, the saturation channel gradient map, and the brightness channel gradient map, respectively, and takes the peak direction of the gradient direction histogram as the local gradient principal direction for that pixel, including: Centered on each pixel in the hue channel gradient map, the saturation channel gradient map, and the brightness channel gradient map, a local neighborhood window with a size of 4N×4N pixels is set, and the local neighborhood window is divided into N×N cells, each cell with a size of N×N pixels, where N is an even number; Within each cell, the gradient direction is quantized into nine intervals, and the cumulative gradient magnitude within each interval is calculated to obtain the nine-dimensional gradient direction histogram of the cell. The histograms of N×N cells are concatenated to obtain a 9N×N dimensional feature vector, which is then normalized using the L2 norm to obtain the directional gradient histogram feature vector of the pixel. The peak direction in the feature vector is taken as the main local gradient direction of the pixel.

[0011] Furthermore, the random sampling consensus algorithm is used to iteratively fit the pixels in the hue channel gradient map, the saturation channel gradient map, and the brightness channel gradient map to form a boundary model. Using the principal direction of the local gradient as a constraint, pixels whose angle between the principal direction of the local gradient and the normal direction of the boundary model is less than a preset angle are determined as interior points. These interior points are then used as the confirmed boundary points. Let P = (The set of candidate boundary points in the hue channel gradient map, the saturation channel gradient map, and the brightness channel gradient map are given by this expression). The i-th pixel Includes its position coordinates ( , and the principal direction of the local gradient ,in For the i-th pixel Horizontal column coordinates, For the i-th pixel Vertical row coordinate, 1≤ i ≤ N ; Set the maximum number of iterations to K, the interior point distance threshold to d, the direction deviation threshold to φ, and the interior point number threshold to T; The minimum sample set is randomly selected from the candidate boundary point set P. The minimum sample set consists of two pixels for the straight line model and three pixels for the curve model. The initial boundary model M is obtained by fitting the pixels in the minimum sample set. Traverse the remaining pixels in the candidate boundary point set P, calculate the distance of each pixel to the initial boundary model M, and calculate the angle between the principal direction of the local gradient of each pixel and the normal direction of the initial boundary model M. If the distance is less than the inlier distance threshold d and the angle is less than the direction deviation threshold φ, then mark the pixel as an inlier. If the number of interior points is greater than the current optimal number of interior points, then update the initial boundary model to obtain the optimal model M_best; The minimum sample set is randomly drawn from the candidate boundary point set P and the initial boundary model is updated repeatedly to obtain the optimal model M_best, until the maximum number of iterations K is reached; Using all the interior points corresponding to the optimal model M_best, the final boundary model is refitted, and all the interior points corresponding to the optimal model M_best are output as the confirmed boundary points. The remaining pixels in the candidate boundary point set P are removed as outliers.

[0012] Furthermore, the step of refitting the final boundary model using all interior points corresponding to the optimal model M_best includes: When the boundary of the closure seam is in the shape of a straight line, the final boundary model is obtained by fitting a straight line model. When the boundary of the closure seam is curved, a quadratic or cubic curve model is used for fitting to obtain the final boundary model.

[0013] Furthermore, the step of planning the working trajectory of the laser rust removal head in the closure joint area based on the confirmed boundary points includes: The boundary points of the hue channel image, saturation channel image, and brightness channel image are fused together. The union of the three sets of boundary points is taken, and the boundary contour lines between the painted area and the ablation area, and between the ablation area and the bare metal area are fitted by the least squares method. Using the boundary outline between the ablation area and the bare metal area as the starting boundary and the boundary outline between the painted area and the ablation area as the ending boundary, a laser rust removal scanning path covering the entire ablation area is generated. The scanning path uses a bow-shaped filling method to perform reciprocating scanning along the vertical direction of the closure joint length, or uses a three-shaped filling method to perform unidirectional scanning along the vertical direction of the closure joint length. Based on the scanning width W of the laser rust removal head and the width D of the ablation area, the overlap rate R between adjacent scanning lines is dynamically adjusted to ensure that the ablation area is completely covered. The formula for calculating the overlap rate R is as follows: ,in For safety factor; The scanning path is converted into work instructions that the robot can execute, and the laser rust removal head is controlled to perform rust removal work along the scanning path.

[0014] This application also provides a laser rust removal operation trajectory planning system for closure seams, the system comprising: The image acquisition module is used to acquire the original image of the seam area and convert the original image from the RGB color space to the HSV color space; The image processing module is used to divide the seam area into a painted area, an ablation area, and a bare metal area in the HSV color space according to preset hue threshold, saturation threshold, and brightness threshold. The gradient map generation module is used to calculate the edge gradients of the hue channel image, saturation channel image, and brightness channel image in the horizontal and vertical directions, respectively, to obtain three sets of independent gradient magnitude and gradient direction information, and generate hue channel gradient maps, saturation channel gradient maps, and brightness channel gradient maps respectively. The direction determination module is used to extract the gradient direction histogram in the local neighborhood of each pixel in the hue channel gradient map, the saturation channel gradient map, and the brightness channel gradient map based on the direction gradient histogram operator, and take the peak direction of the gradient direction histogram as the local gradient main direction of the pixel. The image boundary determination module is used to perform iterative fitting of the pixels in the hue channel gradient map, the saturation channel gradient map, and the brightness channel gradient map to form a boundary model using a random sampling consensus algorithm. With the main local gradient direction as a constraint, pixels whose angle between the main local gradient direction and the normal direction of the boundary model is less than a preset angle are determined as interior points, and the interior points are used as the confirmed boundary points. The trajectory planning module is used to plan the working trajectory of the laser rust removal head in the closure joint area based on the confirmed boundary points.

[0015] This application also provides a computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, it implements the steps of any of the above-described laser rust removal operation trajectory planning methods for closure seams.

[0016] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of any of the above-described laser rust removal operation trajectory planning methods for closure seams.

[0017] This application improves the accuracy and robustness of region classification in complex operating environments by converting the original image of the seam area from RGB color space to HSV color space and accurately dividing the painted, ablated, and bare metal areas according to preset thresholds. Simultaneously, by calculating the edge gradients of the three channels, extracting the principal directions of local gradients, and combining them with a random sampling consensus algorithm to achieve robust fitting and filtering of boundary points, interference factors such as noise and false edges can be eliminated, obtaining continuous and reliable region boundaries, significantly improving the shortcomings of insufficient accuracy in single visual feature boundary positioning. Finally, based on the precise boundary planning, the laser rust removal operation trajectory can achieve complete coverage and precise rust removal of the ablated area, avoiding problems such as incomplete rust removal, paint residue, or substrate damage caused by trajectory deviation, greatly improving the trajectory control accuracy, operational stability, and rust removal quality of laser rust removal operations on ship seams. Attached Figure Description

[0018] To more clearly illustrate the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart illustrating the laser rust removal operation trajectory planning method for closure seams in one embodiment of this application; Figure 2 This is a schematic diagram of HSV color space region division in one embodiment of this application; Figure 3 This is a structural block diagram of a laser rust removal operation trajectory planning system for closure seams in one embodiment of this application; Figure 4 This is a schematic diagram of the interactive process of the laser rust removal operation trajectory planning system for closure seams in one embodiment of this application; Figure 5 This is an internal structural diagram of a computer device in one embodiment of this application. Detailed Implementation

[0020] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the protection scope of this application.

[0021] It should be noted that, in the description of this application, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. The terms "first," "second," etc., in this application are used to distinguish similar objects and are not used to describe a specific order or sequence.

[0022] To enable those skilled in the art to better understand the present application, the present application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0023] like Figure 1 As shown, an embodiment of this application provides a method for planning the trajectory of laser rust removal operation at a closure joint, including the following steps: S1. Obtain the original image of the seam area and convert the original image from RGB color space to HSV color space; S2. In the HSV color space, the closure seam area is divided into a painted area, an ablation area and a bare metal area according to the preset hue threshold, saturation threshold and brightness threshold. S3. Calculate the edge gradients of the hue channel image, saturation channel image, and brightness channel image in the horizontal and vertical directions respectively, obtain three sets of independent gradient magnitude and gradient direction information, and generate hue channel gradient map, saturation channel gradient map, and brightness channel gradient map respectively. S4. Based on the gradient direction histogram operator, extract the gradient direction histogram in the local neighborhood for each pixel in the hue channel gradient map, the saturation channel gradient map, and the brightness channel gradient map, and take the peak direction of the gradient direction histogram as the local gradient main direction of the pixel. S5. Using a random sampling consensus algorithm, the pixels in the hue channel gradient map, the saturation channel gradient map, and the brightness channel gradient map are iteratively fitted to form a boundary model. With the main local gradient direction as a constraint, the pixels whose angle between the main local gradient direction and the normal direction of the boundary model is less than a preset angle are determined as interior points, and the interior points are used as the confirmed boundary points. S6. Based on the confirmed boundary points, plan the working trajectory of the laser rust removal head in the closure joint area.

[0024] Among them, the closure seam refers to the welded area at the joint of the hull sections, which is a key part that needs to be rust-removed during ship construction.

[0025] Laser rust removal is a new type of cleaning process that uses a high-energy laser beam to act on the surface rust layer, causing the contaminants to be vaporized or peeled off instantly. It has the characteristics of being non-contact, having no consumables, and having a small heat-affected zone.

[0026] The HSV color space is a color representation model that includes three dimensions: hue, saturation, and value. These three channels are independent of each other and can better simulate the human eye's perception of color.

[0027] Histogram of Oriented Gradients (HOG) is a feature descriptor used in computer vision and image processing. It characterizes the local shape features of an object by statistically analyzing the distribution of gradient directions in local regions of an image.

[0028] The Random Sample Consensus (RANSAC) algorithm is an iterative parameter estimation method that can robustly estimate mathematical model parameters from datasets containing a large number of outliers.

[0029] By converting the original image of the closure seam from RGB to HSV color space and dividing it into painted, ablated, and bare metal areas according to thresholds, and then calculating the three-channel edge gradient and extracting the main direction of the local gradient, combined with the random sampling consensus algorithm to accurately select boundary points and plan the operation trajectory, it can stably distinguish different areas and accurately locate boundaries in complex scenarios with uneven lighting, rust, and oil stains. This effectively improves the accuracy of laser rust removal trajectory planning, avoids incomplete rust removal or damage to the substrate, and ensures the stability and reliability of the operation.

[0030] like Figure 2As shown, in the HSV color space, based on preset hue thresholds, saturation thresholds, and lightness thresholds, the seam area is divided into a painted area, an ablation area, and a bare metal area, including: A first hue threshold is set based on the difference in hue between the paint layer and the metal substrate, and pixels whose hue values ​​fall within the range of the first hue threshold are divided into paint areas. A saturation threshold is set based on the difference in saturation between bare metal and paint layer, and pixels with saturation values ​​lower than the saturation threshold are classified as bare metal regions. A brightness threshold is set based on the difference in brightness between the ablation and discoloration areas and the original painted areas. Pixels with brightness values ​​lower than the brightness threshold and not belonging to the painted areas are classified as ablation areas.

[0031] Specifically, by setting thresholds based on differences in hue, saturation, and brightness to divide painted, bare metal, and ablation areas, the independent characteristics of each channel in the HSV color space are fully utilized to accurately distinguish the color and brightness characteristics of the three types of areas, reduce the influence of lighting and interference factors, improve the accuracy of area segmentation, and provide a reliable foundation for subsequent boundary extraction and trajectory planning.

[0032] Optionally, for the blue epoxy primer commonly used in ship joints, its hue value is typically in the range of 200°-240°, with a high saturation value (usually greater than 0.5) and a medium lightness value. In ablation areas, due to laser action causing carbonization and discoloration of the paint layer, the hue value shifts to the range of 30°-60°, with a significantly reduced saturation (usually less than 0.3) and a lower lightness value. The bare metal area appears silver-gray, with no obvious hue characteristics, extremely low saturation (usually less than 0.1), and a medium-to-high lightness value. Based on these characteristics, the following typical thresholds can be set: Painted areas: H∈200°, 240°, S∈0.5, 1.0, V∈0.4, 0.8; Ablated areas: H∈30°, 60°, S∈0, 0.3, V∈0.2, 0.5; Bare metal areas: S∈0, 0.1, V∈0.5, 1.0. In actual operation, the above thresholds can be adaptively adjusted according to the specific coating type and laser process parameters.

[0033] Furthermore, the edge gradients of the hue channel image, saturation channel image, and brightness channel image are calculated in the horizontal and vertical directions respectively, resulting in three sets of independent gradient magnitude and gradient direction information, including: The Sobel operator is used to calculate the gradient magnitudes of the hue channel image, saturation channel image, and brightness channel image in the horizontal and vertical directions, respectively. Let the grayscale value of the pixel at position (x,y) in any of the three channels of the image be I_H(x,y), where x is the horizontal column coordinate of the image pixel and y is the vertical row coordinate of the image pixel; The formulas for calculating the horizontal gradient G_x at pixel (x,y) and the vertical gradient G_y at pixel (x,y) are as follows: G_x = I_H(x+1,y-1) + 2I_H(x+1,y) + I_H(x+1,y+1) - I_H(x-1,y-1) - 2I_H(x-1,y) - I_H(x-1,y+1); where I_H(x+1,y-1) is the pixel grayscale value at the position (x+1,y-1), I_H(x+1,y) is the pixel grayscale value at the position (x+1,y+1), I_H(x-1,y-1) is the pixel grayscale value at the position (x+1,y+1), I_H(x-1,y-1) is the pixel grayscale value at the position (x-1,y+1), I_H(x-1,y) is the pixel grayscale value at the position (x-1,y+1), I_H(x-1,y+1) is the pixel grayscale value at the position (x-1,y+1); G_y = I_H(x-1,y+1) + 2I_H(x,y+1) + I_H(x+1,y+1) - I_H(x-1,y-1) - 2I_H(x,y-1) - I_H(x+1,y-1); where I_H(x-1,y+1) is the pixel grayscale value corresponding to the pixel position (x-1,y+1), I_H(x,y+1) is the pixel grayscale value corresponding to the pixel position (x,y+1), I_H(x+1,y+1) is the pixel grayscale value corresponding to the pixel position (x+1,y+1), I_H(x-1,y-1) is the pixel grayscale value corresponding to the pixel position (x-1,y-1), I_H(x,y-1) is the pixel grayscale value corresponding to the pixel position (x,y-1), I_H(x+1,y-1) is the pixel grayscale value corresponding to the pixel position (x+1,y-1); The formulas for calculating the gradient magnitude M(x,y) and gradient direction θ(x,y) are as follows: M(x,y) = ; θ(x,y) = ; The gradient magnitude matrix and gradient direction matrix corresponding to the hue channel image, the saturation channel image, and the brightness channel image are obtained.

[0034] The Sobel operator is used to calculate the horizontal and vertical edge gradients of the three HSV channels and obtain the gradient magnitude and direction matrix. The edge information of each channel is extracted independently to avoid the loss of key boundary features by a single channel gradient, thereby enhancing the integrity and anti-interference ability of edge detection and providing accurate data support for gradient direction analysis and boundary point selection.

[0035] Furthermore, the gradient direction histogram operator, based on the gradient direction histogram, extracts the gradient direction histogram within the local neighborhood for each pixel in the hue channel gradient map, the saturation channel gradient map, and the brightness channel gradient map, respectively, and takes the peak direction of the gradient direction histogram as the local gradient principal direction for that pixel, including: Centered on each pixel in the hue channel gradient map, the saturation channel gradient map, and the brightness channel gradient map, a local neighborhood window with a size of 4N×4N pixels is set, and the local neighborhood window is divided into N×N cells, each cell with a size of N×N pixels, where N is an even number; Within each cell, the gradient direction is quantized into nine intervals, and the cumulative gradient magnitude within each interval is calculated to obtain the nine-dimensional gradient direction histogram of the cell. The histograms of N×N cells are concatenated to obtain a 9N×N dimensional feature vector, which is then normalized using the L2 norm to obtain the directional gradient histogram feature vector of the pixel. The peak direction in the feature vector is taken as the main local gradient direction of the pixel.

[0036] Specifically, the local neighborhood gradient direction histogram of each pixel is extracted based on the directional gradient histogram operator, and the main direction of the local gradient is determined. The edge direction is stably represented by local window statistics and normalization processing, which weakens noise and local abrupt interference, provides consistent and reliable directional constraints for boundary fitting, and improves the accuracy of boundary direction judgment.

[0037] Optionally, the size of the local neighborhood window can be adjusted according to the image resolution and the actual edge scale. Preferably, a local neighborhood window of 16×16 pixels is set, and the window is divided into 4×4 cells, each cell being 4×4 pixels in size. For high-resolution images, a 32×32 pixel window can be used to obtain more macroscopic orientation statistics; for low-resolution images, an 8×8 pixel window can be used to preserve local details.

[0038] Furthermore, the random sampling consensus algorithm is used to iteratively fit the pixels in the hue channel gradient map, the saturation channel gradient map, and the brightness channel gradient map to form a boundary model. Using the principal direction of the local gradient as a constraint, pixels whose angle between the principal direction of the local gradient and the normal direction of the boundary model is less than a preset angle are determined as interior points. These interior points are then used as the confirmed boundary points, including: Let P = (The set of candidate boundary points in the hue channel gradient map, the saturation channel gradient map, and the brightness channel gradient map are given by this expression). The i-th pixel Includes its position coordinates ( , and the principal direction of the local gradient ,in For the i-th pixel Horizontal column coordinates, For the i-th pixel Vertical row coordinate, 1≤ i ≤ N ; Set the maximum number of iterations to K (usually 500-1000 times), the interior point distance threshold to d (usually 2-5 pixels), the orientation deviation threshold to φ (usually 10°-20°), and the interior point number threshold to T; The minimum sample set is randomly selected from the candidate boundary point set P. The minimum sample set consists of two pixels for the straight line model and three pixels for the curve model. The initial boundary model M is obtained by fitting the pixels in the minimum sample set. Traverse the remaining pixels in the candidate boundary point set P, calculate the distance of each pixel to the initial boundary model M, and calculate the angle between the principal direction of the local gradient of each pixel and the normal direction of the initial boundary model M. If the distance is less than the inlier distance threshold d and the angle is less than the direction deviation threshold φ, then mark the pixel as an inlier. If the number of interior points is greater than the current optimal number of interior points, then update the initial boundary model to obtain the optimal model M_best; The minimum sample set is randomly drawn from the candidate boundary point set P and the initial boundary model is updated repeatedly to obtain the optimal model M_best, until the maximum number of iterations K is reached; Using all the interior points corresponding to the optimal model M_best, the final boundary model is refitted, and all the interior points corresponding to the optimal model M_best are output as the confirmed boundary points. The remaining pixels in the candidate boundary point set P are removed as outliers.

[0039] Among them, the boundary model is iteratively fitted by a random sampling consensus algorithm with the main direction of local gradient as a constraint and the interior points are selected as valid boundary points. This can automatically remove outliers such as noise and pseudo-edges, and robustly fit a continuous and accurate regional boundary, which greatly improves the boundary positioning accuracy and anti-interference ability.

[0040] Furthermore, the step of refitting the final boundary model using all interior points corresponding to the optimal model M_best includes: When the boundary of the closure seam is in the shape of a straight line, the final boundary model is obtained by fitting a straight line model. When the boundary of the closure seam is curved, a quadratic or cubic curve model is used for fitting to obtain the final boundary model.

[0041] Among them, the corresponding model is selected to refit the optimal inner point according to the straight or curved shape of the closure joint boundary, adapting to the straight and complex curved closure joint boundaries, ensuring that the boundary fitting fits the actual shape, further improving the boundary contour accuracy, and adapting to rust removal scenarios of closure joints with different structures.

[0042] Furthermore, the step of planning the working trajectory of the laser rust removal head in the closure joint area based on the confirmed boundary points includes: The boundary points of the hue channel image, saturation channel image, and brightness channel image are fused together. The union of the three sets of boundary points is taken, and the boundary contour lines between the painted area and the ablation area, and between the ablation area and the bare metal area are fitted by the least squares method. Using the boundary outline between the ablation area and the bare metal area as the starting boundary and the boundary outline between the painted area and the ablation area as the ending boundary, a laser rust removal scanning path covering the entire ablation area is generated. The scanning path uses a bow-shaped filling method to perform reciprocating scanning along the vertical direction of the closure joint length, or uses a three-shaped filling method to perform unidirectional scanning along the vertical direction of the closure joint length. Based on the scanning width W of the laser rust removal head and the width D of the ablation area, the overlap rate R between adjacent scanning lines is dynamically adjusted to ensure that the ablation area is completely covered. The formula for calculating the overlap rate R is as follows: ,in For safety factor; The scanning path is converted into work instructions that the robot can execute, and the laser rust removal head is controlled to perform rust removal work along the scanning path.

[0043] The value of η ranges from 0.8 to 0.95 to ensure that the ablation area can still be completely covered even with boundary positioning errors.

[0044] Optionally, when the joint length is long, the scanning path can be segmented, with each segment not exceeding the working stroke of the laser rust removal head, and overlapping areas set at the segmentation points to ensure that no joint is missed.

[0045] By fusing the boundary points of the three channels and fitting the dividing contour line, a bow-shaped or triangular scanning path is generated and the scanning overlap rate is dynamically adjusted. This can completely cover the ablation area, avoiding missed rust removal and repeated excessive irradiation. At the same time, the path is converted into robot-executable instructions to achieve automated and precise rust removal, taking into account both rust removal quality and work efficiency.

[0046] This invention utilizes multi-threshold segmentation in the HSV color space to fully leverage information from three dimensions: hue, saturation, and brightness. This effectively distinguishes between painted areas, ablated areas, and bare metal areas, overcoming the shortcomings of single grayscale features or the susceptibility of RGB space to lighting effects, and improving the accuracy and robustness of area classification.

[0047] This invention calculates the independent gradients of the three channels of hue, saturation, and brightness, and uses the directional gradient histogram operator to extract the main direction of the local gradient. By using a random sampling consensus algorithm to vote on the gradient direction, it can effectively eliminate false edge points caused by noise, rust, oil stains, etc., and obtain continuous and accurate regional boundary points, significantly improving the accuracy and reliability of boundary positioning.

[0048] This invention plans the laser rust removal operation trajectory based on accurately identified area boundaries, which can achieve precise coverage of the ablation area, avoid damage to painted areas and excessive irradiation of bare metal areas, and support dynamic adjustment of the scan line overlap rate. It improves operation efficiency while ensuring rust removal quality and is suitable for automated laser rust removal operations in complex ship closure joint scenarios.

[0049] In the aforementioned laser rust removal trajectory planning method for closure seams, the original image of the closure seam area is converted from RGB color space to HSV color space, and the painted, ablated, and bare metal areas are accurately divided according to a preset threshold. This improves the accuracy and robustness of area classification in complex operating environments. Simultaneously, by calculating the edge gradients of the three channels, extracting the principal directions of local gradients, and combining them with a random sampling consensus algorithm, robust fitting and filtering of boundary points are achieved. This eliminates interference factors such as noise and false edges, obtaining continuous and reliable area boundaries, significantly improving the shortcomings of insufficient positioning accuracy of single visual feature boundaries. Finally, based on the precise boundary planning, the laser rust removal trajectory can achieve complete coverage and precise rust removal of the ablated area, avoiding problems such as incomplete rust removal, paint residue, or substrate damage caused by trajectory deviation. This significantly improves the trajectory control accuracy, operational stability, and rust removal quality of laser rust removal operations for ship closure seams, meeting the practical application requirements of automated laser rust removal.

[0050] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods according to the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method.

[0051] In one embodiment, such as Figure 3 , Figure 4 As shown, a laser rust removal operation trajectory planning system 10 for closure seams is provided, including: an image acquisition module 1, an image processing module 2, a gradient map generation module 3, a direction determination module 4, an image boundary determination module 5, and a trajectory planning module 6.

[0052] Image acquisition module 1 is used to acquire the original image of the seam area and convert the original image from RGB color space to HSV color space; Image processing module 2 is used to divide the seam area into a painted area, an ablation area and a bare metal area in the HSV color space according to preset hue threshold, saturation threshold and brightness threshold. The gradient map generation module 3 is used to calculate the edge gradients of the hue channel image, saturation channel image, and brightness channel image in the horizontal and vertical directions respectively, obtain three sets of independent gradient magnitude and gradient direction information, and generate hue channel gradient map, saturation channel gradient map, and brightness channel gradient map respectively. The direction determination module 4 is used to extract the gradient direction histogram in the local neighborhood of each pixel in the hue channel gradient map, the saturation channel gradient map, and the brightness channel gradient map based on the direction gradient histogram operator, and take the peak direction of the gradient direction histogram as the local gradient main direction of the pixel. The image boundary determination module 5 is used to perform iterative fitting of the pixels in the hue channel gradient map, the saturation channel gradient map, and the brightness channel gradient map to form a boundary model using a random sampling consensus algorithm. With the main local gradient direction as a constraint, the pixels with an angle less than a preset angle between the main local gradient direction and the normal direction of the boundary model are determined as interior points, and the interior points are used as the confirmed boundary points. The trajectory planning module 6 is used to plan the working trajectory of the laser rust removal head in the closure joint area based on the confirmed boundary points.

[0053] The aforementioned laser rust removal trajectory planning system for seams improves the accuracy and robustness of region classification in complex operating environments by converting the original image of the seam area from RGB color space to HSV color space and accurately dividing the painted, ablated, and bare metal areas according to preset thresholds. Simultaneously, by calculating the edge gradients of the three channels, extracting the principal directions of local gradients, and combining them with a random sampling consensus algorithm, robust fitting and filtering of boundary points are achieved. This eliminates interference factors such as noise and false edges, obtaining continuous and reliable region boundaries, significantly improving the insufficient accuracy of single visual feature boundary positioning. Finally, based on precise boundary planning, the laser rust removal trajectory can achieve complete coverage and precise rust removal of the ablated area, avoiding incomplete rust removal, paint residue, or substrate damage caused by trajectory deviation. This significantly improves the trajectory control accuracy, operational stability, and rust removal quality of laser rust removal operations on ship seams, meeting the practical application requirements of automated laser rust removal.

[0054] For a description of the features in the embodiment of the laser rust removal operation trajectory planning system for closure seams, please refer to the relevant description of the embodiment of the laser rust removal operation trajectory planning method for closure seams, which will not be repeated here.

[0055] Embodiments of this application also provide an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to perform the steps in any of the embodiments of the laser rust removal operation trajectory planning method for closure seams described above.

[0056] In one embodiment, the electronic device may be a server, and its internal structure diagram may be as follows: Figure 5 As shown, the electronic device includes a processor, memory, network interface, and database connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores the operating system, computer programs, and database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database stores trajectory planning data for laser rust removal operations at closure seams. The network interface communicates with external terminals via a network connection. When the computer program is executed by the processor, it implements a method for planning laser rust removal operation trajectories at closure seams.

[0057] Embodiments of this application also provide a computer-readable storage medium storing a computer program, wherein the computer program is configured to execute the steps in any of the above embodiments of the laser rust removal operation trajectory planning method for closure seams when it is run.

[0058] In one exemplary embodiment, the aforementioned computer-readable storage medium may include, but is not limited to, various media capable of storing computer programs, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard disk, magnetic disk, or optical disk.

[0059] The embodiments of this application also provide a computer program product, which includes a computer program that, when executed by a processor, implements the steps in any of the embodiments of the laser rust removal operation trajectory planning method for closure seams described above.

[0060] Embodiments of this application also provide another computer program product, including a non-volatile computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps in any of the embodiments of the laser rust removal operation trajectory planning method for closure seams described above.

[0061] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0062] The foregoing has provided a detailed description of the laser rust removal trajectory planning method, system, medium, and product for closure seams provided in this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the embodiments above are only intended to aid in understanding the method and core ideas of this application. It should be noted that those skilled in the art can make various improvements and modifications to this application without departing from its principles, and these improvements and modifications also fall within the protection scope of the claims of this application.

Claims

1. A method for planning the trajectory of laser rust removal operation at closure seams, characterized in that, include: Obtain the original image of the seam area and convert the original image from the RGB color space to the HSV color space; In the HSV color space, the seam area is divided into a painted area, an ablation area, and a bare metal area according to preset hue threshold, saturation threshold, and brightness threshold. The edge gradients of the hue channel image, saturation channel image, and brightness channel image in the horizontal and vertical directions are calculated respectively to obtain three sets of independent gradient magnitude and gradient direction information, and hue channel gradient map, saturation channel gradient map, and brightness channel gradient map are generated respectively. Based on the directional gradient histogram operator, the gradient direction histogram in the local neighborhood is extracted for each pixel in the hue channel gradient map, the saturation channel gradient map, and the brightness channel gradient map, respectively, and the peak direction of the gradient direction histogram is taken as the local gradient principal direction of the pixel. A random sampling consensus algorithm is used to iteratively fit the pixels in the hue channel gradient map, the saturation channel gradient map, and the brightness channel gradient map to form a boundary model. With the main local gradient direction as a constraint, pixels whose angle between the main local gradient direction and the normal direction of the boundary model is less than a preset angle are determined as interior points, and the interior points are used as the confirmed boundary points. Based on the confirmed boundary points, plan the working trajectory of the laser rust removal head in the closure joint area.

2. The laser rust removal operation trajectory planning method for closure seams according to claim 1, characterized in that, In the HSV color space, based on preset hue thresholds, saturation thresholds, and brightness thresholds, the seam area is divided into a painted area, an ablation area, and a bare metal area, including: A first hue threshold is set based on the difference in hue between the paint layer and the metal substrate, and pixels whose hue values ​​fall within the range of the first hue threshold are divided into paint areas. A saturation threshold is set based on the difference in saturation between bare metal and paint layer, and pixels with saturation values ​​lower than the saturation threshold are classified as bare metal regions. A brightness threshold is set based on the difference in brightness between the ablation and discoloration areas and the original painted areas. Pixels with brightness values ​​lower than the brightness threshold and not belonging to the painted areas are classified as ablation areas.

3. The laser rust removal operation trajectory planning method for closure seams according to claim 1, characterized in that, The process involves calculating the edge gradients of the hue channel image, saturation channel image, and brightness channel image in the horizontal and vertical directions, respectively, to obtain three sets of independent gradient magnitude and gradient direction information, including: The Sobel operator is used to calculate the gradient magnitudes of the hue channel image, saturation channel image, and brightness channel image in the horizontal and vertical directions, respectively. Let the grayscale value of the pixel at position (x,y) in any of the three channels of the image be I_H(x,y), where x is the horizontal column coordinate of the image pixel and y is the vertical row coordinate of the image pixel; The formulas for calculating the horizontal gradient G_x at pixel (x,y) and the vertical gradient G_y at pixel (x,y) are as follows: G_x = I_H(x+1,y-1) + 2I_H(x+1,y) + I_H(x+1,y+1) - I_H(x-1,y-1) - 2I_H(x-1,y) - I_H(x-1,y+1); where I_H(x+1,y-1) is the pixel grayscale value at the position (x+1,y-1), I_H(x+1,y) is the pixel grayscale value at the position (x+1,y+1), I_H(x-1,y-1) is the pixel grayscale value at the position (x+1,y+1), I_H(x-1,y-1) is the pixel grayscale value at the position (x-1,y+1), I_H(x-1,y) is the pixel grayscale value at the position (x-1,y+1), I_H(x-1,y+1) is the pixel grayscale value at the position (x-1,y+1); G_y = I_H(x-1,y+1) + 2I_H(x,y+1) + I_H(x+1,y+1) - I_H(x-1,y-1) - 2I_H(x,y-1) - I_H(x+1,y-1); where I_H(x-1,y+1) is the pixel grayscale value corresponding to the pixel position (x-1,y+1), I_H(x,y+1) is the pixel grayscale value corresponding to the pixel position (x,y+1), I_H(x+1,y+1) is the pixel grayscale value corresponding to the pixel position (x+1,y+1), I_H(x-1,y-1) is the pixel grayscale value corresponding to the pixel position (x-1,y-1), I_H(x,y-1) is the pixel grayscale value corresponding to the pixel position (x,y-1), I_H(x+1,y-1) is the pixel grayscale value corresponding to the pixel position (x+1,y-1); The formulas for calculating the gradient magnitude M(x,y) and gradient direction θ(x,y) are as follows: M(x,y) = ; θ(x,y) = ; The gradient magnitude matrix and gradient direction matrix corresponding to the hue channel image, the saturation channel image, and the brightness channel image are obtained.

4. The laser rust removal operation trajectory planning method for closure seams according to claim 1, characterized in that, The gradient direction histogram operator extracts the gradient direction histogram within the local neighborhood of each pixel in the hue channel gradient map, the saturation channel gradient map, and the brightness channel gradient map, and takes the peak direction of the gradient direction histogram as the local gradient principal direction of that pixel, including: Centered on each pixel in the hue channel gradient map, the saturation channel gradient map, and the brightness channel gradient map, a local neighborhood window with a size of 4N×4N pixels is set, and the local neighborhood window is divided into N×N cells, each cell with a size of N×N pixels, where N is an even number; Within each cell, the gradient direction is quantized into nine intervals, and the cumulative gradient magnitude within each interval is calculated to obtain the nine-dimensional gradient direction histogram of the cell. The histograms of N×N cells are concatenated to obtain a 9N×N dimensional feature vector, which is then normalized using the L2 norm to obtain the directional gradient histogram feature vector of the pixel. The peak direction in the feature vector is taken as the main local gradient direction of the pixel.

5. The laser rust removal operation trajectory planning method for closure seams according to claim 1, characterized in that, The method employs a random sampling consensus algorithm to iteratively fit the pixels in the hue channel gradient map, saturation channel gradient map, and brightness channel gradient map to form a boundary model. Using the principal direction of the local gradient as a constraint, pixels whose angle between the principal direction of the local gradient and the normal direction of the boundary model is less than a preset angle are identified as interior points. These interior points are then used as the confirmed boundary points. Let P = (The set of candidate boundary points in the hue channel gradient map, the saturation channel gradient map, and the brightness channel gradient map are given by this expression). The i-th pixel Includes its position coordinates ( , and the principal direction of the local gradient ,in For the i-th pixel Horizontal column coordinates, For the i-th pixel Vertical row coordinate, 1≤ i ≤ N ; Set the maximum number of iterations to K, the interior point distance threshold to d, the direction deviation threshold to φ, and the interior point number threshold to T; The minimum sample set is randomly selected from the candidate boundary point set P. The minimum sample set consists of two pixels for the straight line model and three pixels for the curve model. The initial boundary model M is obtained by fitting the pixels in the minimum sample set. Traverse the remaining pixels in the candidate boundary point set P, calculate the distance of each pixel to the initial boundary model M, and calculate the angle between the principal direction of the local gradient of each pixel and the normal direction of the initial boundary model M. If the distance is less than the inlier distance threshold d and the angle is less than the direction deviation threshold φ, then mark the pixel as an inlier. If the number of interior points is greater than the current optimal number of interior points, then update the initial boundary model to obtain the optimal model M_best; The minimum sample set is randomly drawn from the candidate boundary point set P and the initial boundary model is updated repeatedly to obtain the optimal model M_best, until the maximum number of iterations K is reached; Using all the interior points corresponding to the optimal model M_best, the final boundary model is refitted, and all the interior points corresponding to the optimal model M_best are output as the confirmed boundary points. The remaining pixels in the candidate boundary point set P are removed as outliers.

6. The laser rust removal operation trajectory planning method for closure seams according to claim 5, characterized in that, The step of refitting the final boundary model using all interior points corresponding to the optimal model M_best includes: When the boundary of the closure seam is in the shape of a straight line, the final boundary model is obtained by fitting a straight line model. When the boundary of the closure seam is curved, a quadratic or cubic curve model is used for fitting to obtain the final boundary model.

7. The laser rust removal operation trajectory planning method for closure seams according to claim 1, characterized in that, The step of planning the working trajectory of the laser rust removal head in the closure joint area based on the confirmed boundary points includes: The boundary points of the hue channel image, saturation channel image, and brightness channel image are fused together. The union of the three sets of boundary points is taken, and the boundary contour lines between the painted area and the ablation area, and between the ablation area and the bare metal area are fitted by the least squares method. Using the boundary outline between the ablation area and the bare metal area as the starting boundary and the boundary outline between the painted area and the ablation area as the ending boundary, a laser rust removal scanning path covering the entire ablation area is generated. The scanning path uses a bow-shaped filling method to perform reciprocating scanning along the vertical direction of the closure joint length, or uses a three-shaped filling method to perform unidirectional scanning along the vertical direction of the closure joint length. Based on the scanning width W of the laser rust removal head and the width D of the ablation area, the overlap rate R between adjacent scanning lines is dynamically adjusted to ensure that the ablation area is completely covered. The formula for calculating the overlap rate R is as follows: ,in For safety factor; The scanning path is converted into work instructions that the robot can execute, and the laser rust removal head is controlled to perform rust removal work along the scanning path.

8. A laser rust removal operation trajectory planning system for closure seams, characterized in that, The system includes: The image acquisition module is used to acquire the original image of the seam area and convert the original image from the RGB color space to the HSV color space; The image processing module is used to divide the seam area into a painted area, an ablation area, and a bare metal area in the HSV color space according to preset hue threshold, saturation threshold, and brightness threshold. The gradient map generation module is used to calculate the edge gradients of the hue channel image, saturation channel image, and brightness channel image in the horizontal and vertical directions, respectively, to obtain three sets of independent gradient magnitude and gradient direction information, and generate hue channel gradient maps, saturation channel gradient maps, and brightness channel gradient maps respectively. The direction determination module is used to extract the gradient direction histogram in the local neighborhood of each pixel in the hue channel gradient map, the saturation channel gradient map, and the brightness channel gradient map based on the direction gradient histogram operator, and take the peak direction of the gradient direction histogram as the local gradient main direction of the pixel. The image boundary determination module is used to perform iterative fitting of the pixels in the hue channel gradient map, the saturation channel gradient map, and the brightness channel gradient map to form a boundary model using a random sampling consensus algorithm. With the main local gradient direction as a constraint, pixels whose angle between the main local gradient direction and the normal direction of the boundary model is less than a preset angle are determined as interior points, and the interior points are used as the confirmed boundary points. The trajectory planning module is used to plan the working trajectory of the laser rust removal head in the closure joint area based on the confirmed boundary points.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, wherein when the computer program is executed by a processor, it implements the steps of the laser rust removal operation trajectory planning method for closure seams as described in any one of claims 1 to 7.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the laser rust removal operation trajectory planning method for closure seams as described in any one of claims 1 to 7.

Images