A laser vision-based real-time tracking method and system for spatially discontinuous welds

Abstract

The application discloses a kind of real-time tracking method and system of space discontinuous weld based on laser vision, it is related to robot intelligent welding technical field, including steps: real-time acquisition laser stripe image of the region to be welded;Laser stripe image is carried out denoising processing to obtain denoising stripe image, extract image region of interest in denoising stripe image, extract stripe image center point in image region of interest using gray centroid algorithm;Based on stripe image center point, segmented fitting is carried out, and first fitting straight line segment and second fitting straight line segment are obtained;Image intersection point is mapped from image coordinate system to robot coordinate system based on calibration mapping relationship, and predetermined positioning image welding point is identified as welding skip point, and continuous welding center fitted by weld feature image is identified as continuous welding point.The method provided by the application has the beneficial effects of identifying predetermined positioning welding point and continuous welding point in advance to guide welding arm to carry out welding operation, which is beneficial to the automation of welding operation and improves the quality of welding.

Description

Technical Field

[0001] This invention relates to the field of intelligent welding technology, and in particular to a method and system for real-time tracking of spatially discontinuous weld seams based on laser vision. Background Technology

[0002] Welding, as a fundamental structural joining technology, plays a crucial role in various fields such as automobile manufacturing, construction, transportation, and aerospace. With the rapid development of automation technology, welding robots are being used more and more widely, significantly improving the efficiency and consistency of the welding process.

[0003] In practical applications, welding robots are still mainly operated using two modes: "teaching playback" and "offline programming." The "teaching playback" mode relies on manual instruction, requiring operators to manually guide the robot to set the welding path on-site. This method can introduce errors in complex workpieces or variable environments and lacks flexibility. Meanwhile, the "offline programming" mode requires pre-generating the welding path. While this can improve efficiency, for complex welding tasks, the creation and adjustment of the model is often cumbersome and time-consuming.

[0004] For complex three-dimensional structures, pre-tack welding is required before general welding. Failure to weld beyond these pre-tack points during subsequent welding can easily lead to quality issues such as bulging. Currently, common real-time weld seam tracking methods typically rely on sensors, primarily categorized as laser sensors, arc sensors, contact sensors, and vision sensors. First, laser sensors, the most widely used, offer strong anti-interference capabilities, but their complex debugging and high operator skill requirements limit their application on a large scale due to limitations in working distance and the laser triangulation method. Second, arc sensors are insensitive to arcs, temperature, and dust, and do not suffer from signal lead issues; however, their inability to detect weld seam feature points in advance restricts their application in spatially discontinuous weld seams. Third, contact sensors are insufficient for the high-precision requirements of industrial manufacturing. Finally, vision sensors can accurately acquire data in real-time and offer the advantage of advanced data acquisition, making them crucial for tracking spatially discontinuous weld seams in complex working environments. In existing technologies, real-time weld seam tracking methods based on laser vision sensors are only applicable to simple weld seams and cannot meet the real-time tracking requirements of spatially discontinuous weld seams in complex scenarios. Existing line fitting-based techniques can be used for feature point extraction of V-shaped weld seams and fillet weld seams, but most existing technologies rely on traditional image processing techniques to denoise welding images with complex backgrounds and strong noise. Otherwise, high-quality images cannot be obtained for subsequent algorithm analysis, limiting their further development. Therefore, improving the ability of welding robots to detect, identify, and pre-position weld points in complex environments is particularly important. It is necessary to propose a real-time tracking method and system for spatially discontinuous weld seams based on laser vision to solve or at least alleviate some of the above-mentioned shortcomings. Summary of the Invention

[0005] The main objective of this invention is to provide a method and system for real-time tracking of spatially discontinuous welds based on laser vision, aiming to solve the technical problem in the prior art that machine vision is difficult to identify pre-positioned weld points in advance, resulting in repeated welding of pre-positioned weld points and poor welding quality.

[0006] To achieve the above objectives, this invention provides a real-time tracking method for spatially discontinuous weld seams based on laser vision. The method utilizes a front-mounted laser emitter and a front-mounted camera mounted on a welding arm to track weld seams in a spatial three-dimensional structure. The front-mounted laser emitter is located upstream of the welding arm, and the front-mounted camera is also upstream of the laser emitter. The front-mounted camera is tilted at an angle α towards the laser emitter. The spatial three-dimensional structure includes a first plate and a second plate arranged at an angle β, where β is between 0 and 180 degrees. The first and second plates are pre-positioned for welding via pre-positioned weld points. The camera coordinate system corresponding to the front-mounted camera has a calibration mapping relationship with the robot coordinate system of the welding arm. The method includes the following steps:

[0007] S10 captures laser stripe images in real time using the front-facing camera;

[0008] S20, Denoise the laser stripe image to obtain a denoised stripe image, extract the region of interest in the denoised stripe image, and use the gray-scale centroid algorithm to extract the center point of the stripe image in the region of interest.

[0009] S30, based on the center point of the stripe image, the RANSAC straight line fitting algorithm is used to perform segmented fitting to obtain the first and second fitted straight line segments.

[0010] S40. Establish an image coordinate system with the upper left corner of the denoised stripe image as the origin, the vertical extension direction as the x-axis, and the horizontal extension direction as the y-axis. Based on the image coordinate system, determine whether the current slope K1 of the fitted line segment 1 is greater than 0, and determine whether the current slope K2 of the fitted line segment 2 is less than 0.

[0011] S50, if the current slope one K1 is greater than 0 and the current slope two K2 is less than 0, then the intersection point of the images of fitted line segment one and fitted line segment two is obtained as the center point of the weld feature image; if the current slope one K1 is not greater than 0 and the current slope two K2 is less than 0, then the intersection point of the images of fitted line segment one and fitted line segment two is determined as the pre-positioned image weld point.

[0012] S60, based on the calibration mapping relationship, maps the intersection points of the images to the robot coordinate system through the image coordinate system, identifies the pre-positioned image weld points as welding jump points, and identifies the center point of the weld feature image as continuous welding points.

[0013] Furthermore, the pixel coordinates in the image coordinate system ( Mapped to camera coordinates in the camera coordinate system ( The expression for ) is

[0014] ,in, The radial focal length of the camera lens. For the tangential focal length of the camera lens, ( ( ) represents the coordinates of the center point of the laser stripe in the camera coordinate system, obtained through camera photography. The coordinates of the camera's optical center height;

[0015] Using formula Calculate and obtain the coordinates of the center point of the laser stripe ( ),in, , , The equation of the laser plane in the camera coordinate system The plane normal vector;

[0016] Using formula Obtain the calibration mapping relationship between the camera coordinate system and the robot coordinate system (perform hand-eye calibration), whereby... The hand-eye transformation matrix, Let be the transformation matrix of the robot coordinate system, ( ) are the homogeneous coordinates of the laser stripe in the robot coordinate system. () represents the homogeneous coordinates of the laser stripes in the camera coordinate system. The rotation angles along the x, y, and z axes when transforming from world coordinates to camera coordinates. The distance to be translated along the x, y, and z axes from world coordinates to camera coordinates. for.

[0017] Furthermore, step S20 specifically includes:

[0018] The U-NET network model was used to segment the laser stripe image into strong light image and firework image to obtain a denoised stripe image.

[0019] The region of interest in the denoised striped image was determined by the gray-scale mean processing method.

[0020] The gray-scale centroid algorithm is used to extract the center point of the stripes in the region of interest of the image.

[0021] Furthermore, using the formula Determine the region of interest in the image, where, , This represents the row and column coordinates of a point obtained according to its grayscale centroid value in the image coordinate system.

[0022] Furthermore, using the formula Determine the grayscale threshold. It is the maximum gray value in the region of interest of the image. It is the average gray level of the image within the region of interest. It is the grayscale threshold;

[0023] Using formula For the pixels in the i-th column of the image coordinate system The screening process was conducted, among which, Let m be the candidate center point, and m be the corresponding... Pixel value at the point;

[0024] Using formula Get the coordinates of the center point of the i-th column of stripes .

[0025] Further, in step S30, the center point of the stripe image is divided into two sets of fitted coordinate points by a preset cutting block line, wherein the preset cutting block line is parallel to the x-axis and is located at the horizontal midline of the region of interest in the image.

[0026] Furthermore, The value ranges from 40° to 50°.

[0027] Furthermore, β is 90°.

[0028] This invention also provides a real-time tracking system for spatially discontinuous weld seams based on laser vision.

[0029] It includes a welding robot, a front-mounted laser emitter, and a front-mounted camera. The welding robot has a base and a robotic arm, with the front-mounted laser emitter and front-mounted camera both mounted on the robotic arm.

[0030] The welding robot is equipped with a processing device, which is used to implement the steps of the above-mentioned laser vision-based real-time tracking method for spatially discontinuous weld seams.

[0031] Compared with existing technologies, the real-time tracking method for spatially discontinuous welds based on laser vision provided by this invention has the following beneficial effects:

[0032] This invention provides a real-time tracking method for spatially discontinuous welds based on laser vision. The method involves acquiring laser stripe images of the area to be welded in real time using a front-facing camera upstream of the welding arm; denoising the laser stripe images to obtain denoised stripe images; extracting the region of interest (ROI) from the denoised stripe images; and using a grayscale centroid algorithm to extract the center points of the stripe images within the ROI; after segmenting and fitting based on the center points of the stripe images to obtain fitted line segments one and two, determining the intersection points of the images as the center points of the weld feature images or pre-positioned weld points based on the slopes of fitted line segments one and two in the image coordinate system; finally, mapping the image intersection points to the robot coordinate system based on a calibration mapping relationship, identifying the pre-positioned weld points as welding jump points, and identifying the center points of the weld feature images as continuous welding points. The method of the present invention is based on a 2.5D laser vision sensor formed by combining a front-mounted laser emitter and a front-mounted camera. It can acquire the depth information of laser stripes, accurately mark the mapping relationship, and facilitate the segmentation of laser stripe images. Based on the marked mapping relationship and the analysis of the laser stripe images, it can identify pre-positioned welding points and continuous welding points in advance to guide the welding arm to perform welding operations, which is conducive to the automation of welding operations and improves welding quality. Attached Figure Description

[0033] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the structures shown in these drawings without creative effort.

[0034] Figure 1 This is a flowchart illustrating a method for real-time tracking of spatially discontinuous welds based on laser vision, according to one embodiment of the present invention.

[0035] Figure 2 This is a schematic diagram of the mapping features of laser stripes in one embodiment of the present invention, wherein 2a is a schematic diagram of the first correspondence between two laser projection lines when the weld is continuous, and 2b is a schematic diagram of the second correspondence between two laser projection lines when the prepositioned weld point interferes.

[0036] Figure 3 This is a schematic diagram of using the U-NET network model to denoise a laser stripe image in one embodiment of the present invention, wherein 3a is a schematic diagram of strong light image segmentation and 3b is a schematic diagram of fireworks image segmentation;

[0037] Figure 4 This is a schematic diagram illustrating the principle of determining the region of interest in an image according to one embodiment of the present invention;

[0038] Figure 5 This is a schematic diagram of the principle of fitting a first fitting line segment and a second fitting line segment in one embodiment of the present invention. 5a is a schematic diagram of the principle of obtaining continuous welding points, 5b is one of the schematic diagrams of the principle of obtaining welding jump points, and 5c is another schematic diagram of the principle of obtaining welding jump points.

[0039] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0040] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0041] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0042] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present invention are only used to explain the relative positional relationship and movement of each component in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indication will also change accordingly.

[0043] Furthermore, the use of terms such as "first" and "second" in this invention is for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of that feature. Additionally, the technical solutions of the various embodiments can be combined with each other, but only on the basis of being achievable by those skilled in the art. When the combination of technical solutions is contradictory or impossible to implement, such a combination of technical solutions should be considered non-existent and not within the scope of protection claimed by this invention.

[0044] Please refer to the appendix. Figure 1 , Figure 2 , Figure 3 , Figure 4 as well as Figure 5 This invention provides a real-time tracking method for spatially discontinuous weld seams based on laser vision. It utilizes a front-mounted laser emitter and a front-mounted camera mounted on a welding arm to track weld seams in a spatial three-dimensional structure. The front-mounted laser emitter is located upstream of the welding arm, and the front-mounted camera is also upstream of the laser emitter. The front-mounted camera is tilted at an angle α towards the laser emitter. The spatial three-dimensional structure includes a first plate and a second plate arranged at an angle β, where β is between 0 and 180 degrees. The first and second plates are pre-positioned for welding via pre-positioned weld points. The camera coordinate system corresponding to the front-mounted camera has a calibration mapping relationship with the robot coordinate system of the welding arm. The method includes the following steps:

[0045] S10 captures laser stripe images in real time using the front-facing camera;

[0046] S20, Denoise the laser stripe image to obtain a denoised stripe image, extract the region of interest in the denoised stripe image, and use the gray-scale centroid algorithm to extract the center point of the stripe image in the region of interest.

[0047] S30, based on the center point of the stripe image, the RANSAC straight line fitting algorithm is used to perform segmented fitting to obtain the first and second fitted straight line segments.

[0048] S40. Establish an image coordinate system with the upper left corner of the denoised stripe image as the origin, the vertical extension direction as the x-axis, and the horizontal extension direction as the y-axis. Based on the image coordinate system, determine whether the current slope K1 of the fitted line segment 1 is greater than 0, and determine whether the current slope K2 of the fitted line segment 2 is less than 0.

[0049] S50, if the current slope one K1 is greater than 0 and the current slope two K2 is less than 0, then the intersection point of the images of fitted line segment one and fitted line segment two is obtained as the center point of the weld feature image (the center point of the feature image of continuous weld); if the current slope one K1 is not greater than 0 and the current slope two K2 is less than 0, then the intersection point of the images of fitted line segment one and fitted line segment two is determined as the pre-positioned image weld point.

[0050] S60, based on the calibration mapping relationship, maps the intersection points of the images to the robot coordinate system through the image coordinate system, identifies the pre-positioned image weld points as welding jump points, and identifies the center point of the weld feature image as continuous welding points.

[0051] The present invention provides a real-time tracking method for spatially discontinuous welds based on laser vision. This method involves acquiring laser stripe images of the area to be welded in real time using a front-facing camera upstream of the welding arm; denoising the laser stripe images to obtain denoised stripe images; extracting the region of interest (ROI) from the denoised stripe images; and using a grayscale centroid algorithm to extract the center points of the stripe images within the ROI. After segmenting and fitting the stripe image center points to obtain fitted line segments one and two, the intersection points are determined as the center points of the weld feature images or pre-positioned weld points based on the slopes of fitted line segments one and two in the image coordinate system. Finally, based on a calibration mapping relationship, the image intersection points are mapped from the image coordinate system to the robot coordinate system, identifying the pre-positioned weld points as welding jump points and the center points of the weld feature images as continuous welding points. The method of the present invention is based on a 2.5D laser vision sensor formed by combining a front-mounted laser emitter and a front-mounted camera. It can acquire the depth information of laser stripes, accurately mark the mapping relationship, and facilitate the segmentation of laser stripe images. Based on the marked mapping relationship and the analysis of the laser stripe images, it can identify pre-positioned welding points and continuous welding points in advance to guide the welding arm to perform welding operations, which is conducive to the automation of welding operations and improves welding quality.

[0052] Specifically, using the formula Make a judgment.

[0053] Please refer to Figure 2 and Figure 3Research has shown that after capturing the first and second laser projection lines using a 2.5D vision sensor, if there is a continuous weld between the first and second plates, the first and second laser projection lines have a first correspondence; if there is a pre-positioned weld point between the first and second plates, the first and second laser projection lines have a second correspondence under the interference of the pre-positioned weld point.

[0054] Furthermore, the pixel coordinates in the image coordinate system ( Mapped to camera coordinates in the camera coordinate system ( The expression for ) is

[0055] ,in, The radial focal length of the camera lens. For the tangential focal length of the camera lens, ( The coordinates of the center point of the laser stripe in the camera coordinate system are obtained by the camera (i.e., the coordinates of the center point of the laser stripe mapped in the camera coordinate system). The coordinates of the camera's optical center height;

[0056] Using formula Calculate and obtain the coordinates of the center point of the laser stripe ( ),in, , , The equation of the laser plane in the camera coordinate system The plane normal vector;

[0057] Using formula Obtain the calibration mapping relationship between the camera coordinate system and the robot coordinate system (perform hand-eye calibration), whereby... The hand-eye transformation matrix, Let be the transformation matrix of the robot coordinate system, ( ) are the homogeneous coordinates of the laser stripe in the robot coordinate system. () represents the homogeneous coordinates of the laser stripes in the camera coordinate system. The rotation angles along the x, y, and z axes when transforming from world coordinates to camera coordinates. The distance required for the translation of the x, y, and z axes from world coordinates to camera coordinates is given.

[0058] Understandably, in a specific embodiment of the present invention, firstly, the pixel coordinates of the laser stripe image captured by the front-facing camera are converted into coordinates in the camera coordinate system; secondly, the equation of the laser plane emitted by the laser is obtained in the camera coordinate system; finally, the coordinates are solved using an optical vision model. The specific principles include: obtaining the laser plane equation in the camera coordinate system through a calibration plate; since the captured laser stripe image contains the stripe features of the laser, the weld coordinates must simultaneously satisfy the plane equation and the condition of being captured by the camera's optical center; in robot vision, the core task of hand-eye calibration (calibration mapping relationship) is to convert the obtained coordinates in the camera coordinate system into coordinates in the robot base coordinate system. The hand-eye transformation matrix can be solved using quaternions or the Rodriguez rotation formula, and the transformation matrix of the robot base coordinates is solved using the fundamental formulas of robot kinematics.

[0059] Further, step S20 specifically includes: using the U-NET network model to perform strong light image segmentation and fireworks image segmentation on the laser stripe image to obtain a denoised stripe image; determining the region of interest in the denoised stripe image using the gray-level mean processing method; and extracting the stripe image center point in the region of interest using the gray-level centroid algorithm.

[0060] In one specific embodiment of the present invention, a pre-segmentation network model is established for image semantic segmentation. The pre-segmentation network model is trained using training data to perform strong light image segmentation and fireworks image segmentation to obtain a U-NET network model. The U-NET network model only needs to have the ability to segment laser stripes, thus avoiding the problem of insufficient generalization ability of complex models.

[0061] Furthermore, using the formula Determine the region of interest in the image, where, , This represents the row and column coordinates of a point obtained according to its grayscale centroid value in the image coordinate system.

[0062] Furthermore, using the formula Determine the grayscale threshold. It is the maximum gray value in the region of interest of the image. It is the average gray level of the image within the region of interest. It is the grayscale threshold;

[0063] Using formula For the pixels in the i-th column of the image coordinate system The screening process was conducted, among which, Let m be the candidate center point, and m be the corresponding... Pixel value at the point;

[0064] Using formula Get the coordinates of the center point of the i-th column of stripes .

[0065] Specifically, the true coordinates of the weld are contained within the stripe features. Weld feature extraction mainly includes: pre-determining the Region of Interest (ROI) in the image using the grayscale mean; obtaining the center point of the stripe image using a grayscale center extraction algorithm; and then... Perform fitting. Extract the center point of the stripe image. The coordinates are used to determine the slope K of the fitted straight line, and a threshold is applied to identify whether there is a need for skip soldering.

[0066] In actual welding operations, strong noise interference such as sparks and arc light is often present. In the denoised stripe image after noise reduction processing, most of the strong noise has been effectively filtered out. This represents the denoised fringe image. Since the gray values ​​of the laser stripes in the denoised fringe image are relatively large, the region of interest (ROI) can be estimated by the gray-level centroids in the row and column directions. The location of ). In the obtained Internally, based on grayscale threshold The center line of the stripes was obtained using the gray-scale centroid method.

[0067] Specifically, in step S30, after obtaining the center points of the stripe image, a RANSAC-based straight line fitting algorithm is used to remove center point noise and obtain fitted straight line segment one and fitted straight line segment two. First, two different points are randomly selected from the extracted set of center points (stripe image center points). and This forms a straight line. Then, check if the coordinates of the other center points meet the distance threshold condition. The process involves saving points that meet the conditions as candidate points for estimating the centerline equation and recording the number of points that meet the conditions. Finally, this first step is repeated N times, and the group of candidate points with the largest number of points is used to fit the line equation. The formula used is... Determine the center point of the stripe image, where, The distance from the candidate point to be centered to the fitted line is given. The candidate points are to be determined as centers. For the finally determined candidate points, the least squares method is used. Perform linear fitting: Where, ( ) represents the slope and intercept of the fitted linear equation. (i=1,2,….n) are candidate points. ) is the average value. n is the number of candidate points. The object detection algorithm successfully fitted and displayed two straight lines.

[0068] Furthermore, the purpose of the invention is to obtain fitting line segment one and fitting line segment two based on the center point of the stripe image. In order to improve the fitting efficiency and the recognition accuracy of the pre-positioned solder joint, in step S30, the center point of the stripe image is divided into two fitting coordinate point groups by a preset cutting block line. The preset cutting block line is parallel to the x-axis and is located at the horizontal midline of the region of interest in the image.

[0069] Furthermore, the laser at an angle Projected onto the surfaces of board one and board two. The value of α ranges from 40° to 50°. Preferably, α is 45 degrees.

[0070] Furthermore, It is 90°.

[0071] Furthermore, the continuous extension length of the welding jump point is obtained, and the welding arc extinguishing strategy is determined based on the continuous extension length.

[0072] The present invention discloses a real-time tracking method for spatially discontinuous weld seams based on laser vision. The main technical means include: designing and building a 2.5D laser vision sensor; pre-capturing a large number of weld seam images to construct a sufficiently rich dataset, thereby training a U-NET neural network model capable of efficient real-time segmentation; inputting the captured laser stripe images into the trained neural network model for segmentation and noise reduction; based on the segmentation results, proposing a real-time target tracking algorithm to accurately obtain the weld seam center point (stripe image center point) within the region of interest of the image; using the RANSAC algorithm to perform linear fitting on the extracted weld seam center point; determining the intersection of the two fitted weld seam lines as welding feature points, and performing threshold judgment on these points; and based on the fact that the slope K of the straight line can reflect the welding characteristics of spatially discontinuous weld seams, providing a decision-making basis for obstacle jumping in the welding path.

[0073] The present invention also provides a real-time tracking system for spatial discontinuous weld seams based on laser vision, including a welding robot, a front-mounted laser emitter and a front-mounted camera. The welding robot has a base and a robotic arm. The front-mounted laser emitter and the front-mounted camera are both mounted on the robotic arm. A processing device is arranged on the welding robot. The processing device is used to implement the steps of the above-mentioned real-time tracking method for spatial discontinuous weld seams based on laser vision.

[0074] The above embodiments are merely illustrative of several implementation methods of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of this application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A laser vision-based real-time tracking method for spatial discontinuous welds, characterized in that a front laser emitter and a front camera are arranged on a welding arm, the front laser emitter is upstream of the welding arm, the front camera is upstream of the front laser emitter, the front camera is arranged at an angle of a in a direction towards the front laser emitter, a spatial three-dimensional structure includes a first plate and a second plate arranged at an angle of β, β is between 0 degrees and 180 degrees, the first plate and the second plate are welded and positioned by a predetermined positioning weld point, a camera coordinate system corresponding to the front camera has a calibrated mapping relationship with a robot coordinate system of the welding arm, and the method comprises the following steps: S10, real-time acquisition of a laser stripe image by the front camera; S20, denoising processing of the laser stripe image to obtain a denoised stripe image, extraction of an image region of interest in the denoised stripe image, and extraction of a stripe image center point in the image region of interest by a gray-scale centroid algorithm; S30, segmented fitting based on the stripe image center point by a RANSAC straight line fitting algorithm to obtain a first fitted straight line segment and a second fitted straight line segment; S40, establishment of an image coordinate system with a left upper corner end point of the denoised stripe image as an origin, a vertical extension direction as an x-axis, and a horizontal extension direction as a y-axis, judgment of whether a current slope one K1 of the first fitted straight line segment is greater than 0 based on the image coordinate system, and judgment of whether a current slope two K2 of the second fitted straight line segment is less than 0 based on the image coordinate system; S50, if the current slope one K1 is greater than 0 and the current slope two K2 is less than 0, the image intersection point of the first fitted straight line segment and the second fitted straight line segment is obtained as a weld feature image center point; if the current slope one K1 is not greater than 0 and the current slope two K2 is less than 0, the image intersection point of the first fitted straight line segment and the second fitted straight line segment is determined as a positioning image weld point; S60, mapping of the image intersection point from the image coordinate system to the robot coordinate system based on the calibrated mapping relationship, identification of the positioning image weld point as a welding skip point, and identification of the weld feature image center point as a continuous welding point.

2. The laser vision-based real-time tracking method for spatial discontinuous welds according to claim 1, characterized in that 3. The laser vision-based real-time tracking method for spatial discontinuous welds according to claim 1, characterized in that Pixel coordinates in the image coordinate system The expression for mapping camera coordinates to the camera coordinate system is: wherein, is the origin of the image coordinate system, is the radial focal length of the camera lens, is the tangential focal length of the camera lens, is the camera coordinate system center coordinate point of the laser stripe acquired by the camera shooting, is the camera optical center height coordinate; Using formula Calculate and obtain the coordinates of the center point of the laser stripe ( ),in, , , The equation of the laser plane in the camera coordinate system The plane normal vector; The formula is A calibration mapping relationship between the camera coordinate system and the robot coordinate system is acquired, wherein, is a hand-eye transformation matrix, is a transformation matrix of the robot coordinate system, is a laser stripe homogeneous coordinate in the robot coordinate system, is a laser stripe homogeneous coordinate in the camera coordinate system, is a rotation angle of x, y and z three axes of the world coordinate transformed to the camera coordinate, is a distance of translation of x, y and z three axes of the world coordinate transformed to the camera coordinate. Step S20 specifically comprises: U-NET network model is used for strong light image segmentation and firework image segmentation of the laser stripe image to obtain a denoised stripe image; An image region of interest in the denoised stripe image is determined by a gray-scale mean value processing method; A gray-scale centroid algorithm is used to extract a stripe image center point in the image region of interest.

4. The laser vision-based real-time tracking method for spatial discontinuous welds according to claim 3, characterized in that 5. The laser vision-based real-time tracking method for spatial discontinuous welds according to claim 4, characterized in that The formula is determining an image region of interest, wherein, , is the row and column coordinate size of the point obtained according to the gray center value in the image coordinate system. ​ using the formula determining a gray scale threshold value, is the maximum gray scale value of the image within the image region of interest, is the mean gray scale value of the image within the image region of interest, is the gray scale threshold value; Using the formula The pixel points in the i-th column search region in the image coordinate system are filtered m is the pixel value at the corresponding point. The formula is The coordinates of the center point of the i-th column of fringes are obtained , N The total number of pixels of the i-th column that satisfy the condition of being greater than after screening. 6.The laser vision based real-time tracking method of spatial discontinuous weld according to any one of claims 1 to 5, characterized in that, In step S30, the center points of the fringe images are divided into two groups of fitting coordinate points by a preset cutting block line, wherein the preset cutting block line is parallel to the x-axis, and the preset cutting block line is located at the horizontal midline position of the image region of interest. 7.The laser vision based real-time tracking method of spatial discontinuous weld according to claim 1, characterized in that, the range of 40° to 50°. 8.The laser vision based real-time tracking method of spatial discontinuous weld according to claim 1, characterized in that, is 90°. 9.A laser vision based real-time tracking system of spatial discontinuous weld, characterized in that, The system comprises a welding robot, a front laser emitter and a front camera, the welding robot has a base and a robot hand, the front laser emitter and the front camera are arranged on the robot hand, A processing device is arranged on the welding robot, and the processing device is used to realize the steps of the laser vision based real-time tracking method of spatial discontinuous weld according to any one of claims 1 to 8.

Images