CN110814465A - Universal method for automatically extracting welding seam contour - Google Patents
Universal method for automatically extracting welding seam contour Download PDFInfo
- Publication number
- CN110814465A CN110814465A CN201911194751.0A CN201911194751A CN110814465A CN 110814465 A CN110814465 A CN 110814465A CN 201911194751 A CN201911194751 A CN 201911194751A CN 110814465 A CN110814465 A CN 110814465A
- Authority
- CN
- China
- Prior art keywords
- image
- welding
- weld
- slope
- reference image
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B23—MACHINE TOOLS; METAL-WORKING NOT OTHERWISE PROVIDED FOR
- B23K—SOLDERING OR UNSOLDERING; WELDING; CLADDING OR PLATING BY SOLDERING OR WELDING; CUTTING BY APPLYING HEAT LOCALLY, e.g. FLAME CUTTING; WORKING BY LASER BEAM
- B23K9/00—Arc welding or cutting
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B23—MACHINE TOOLS; METAL-WORKING NOT OTHERWISE PROVIDED FOR
- B23K—SOLDERING OR UNSOLDERING; WELDING; CLADDING OR PLATING BY SOLDERING OR WELDING; CUTTING BY APPLYING HEAT LOCALLY, e.g. FLAME CUTTING; WORKING BY LASER BEAM
- B23K9/00—Arc welding or cutting
- B23K9/24—Features related to electrodes
- B23K9/28—Supporting devices for electrodes
- B23K9/287—Supporting devices for electrode holders
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B23—MACHINE TOOLS; METAL-WORKING NOT OTHERWISE PROVIDED FOR
- B23K—SOLDERING OR UNSOLDERING; WELDING; CLADDING OR PLATING BY SOLDERING OR WELDING; CUTTING BY APPLYING HEAT LOCALLY, e.g. FLAME CUTTING; WORKING BY LASER BEAM
- B23K9/00—Arc welding or cutting
- B23K9/32—Accessories
Abstract
The invention discloses a general method for automatically extracting a welding seam outline, which relates to the technical field of welding and comprises the following steps: collecting a welding seam image which is not subjected to arc striking before welding as a reference image, and automatically determining a filtering angle and an interested area of Gabor filtering by using the reference image; collecting a welding image, carrying out self-adaptive Gabor filtering on the welding image to obtain a multi-direction characteristic diagram, and simultaneously obtaining a reference image and an interested area part of the direction characteristic diagram; processing the two interested area parts by using a scale invariant feature transform algorithm to obtain the matching positions of the similar points of the two images; and (3) carrying out self-adaptive local threshold segmentation processing on the region of interest of the multidirectional characteristic diagram to obtain a binary image, and extracting the weld seam contour according to the obtained coordinates of the matching position and the clustering result. The invention is effective to typical butt joint and angle joint joints, can be popularized to the extraction of welding seams of other joint outlines, and is beneficial to the integration of a welding information acquisition and processing system.
Description
Technical Field
The invention relates to the technical field of welding, in particular to a general method for extracting a typical butt joint and angle joint GMAW (gas metal arc welding) automatic welding seam profile of a thick plate under laser vision sensing.
Background
At present, in the GMAW (gas metal arc welding) automatic welding of a robot thick plate, a laser vision sensing system is mostly used for detecting characteristic information of a welding seam profile to guide a welding gun to complete welding seam tracking, so that the online effective extraction of the welding seam profile is the premise of realizing the automatic welding. At present, different welding seam contour extraction methods are proposed by a plurality of researches aiming at the GMAW automatic welding of thick plates with different joints. The methods are provided aiming at the profile appearance of the welding seam under the specific application background, and the application range is not wide. As is well known, developing a set of efficient algorithms to achieve online extraction of a certain weld profile requires considerable cost and is not conducive to popularization. Therefore, the universal welding seam outline extraction method not only can save the welding production cost, but also is beneficial to the popularization of the automatic welding technology, and has practical significance.
Disclosure of Invention
Aiming at the problems, the invention provides a general method for automatically extracting the welding seam outline.
The technical scheme adopted by the invention is as follows: a general method for automatic weld seam profile extraction includes the following steps:
collecting a welding seam image which is not subjected to arc striking before welding as a reference image, and automatically determining a filtering angle and an interested area of Gabor filtering by using the reference image; collecting a welding image, carrying out self-adaptive Gabor filtering on the welding image to obtain a multi-direction characteristic diagram, and simultaneously obtaining a reference image and an interested area part of the direction characteristic diagram; processing the two interested area parts by using a scale invariant feature transform algorithm to obtain the matching positions of the similar points of the two images; and (3) carrying out self-adaptive local threshold segmentation processing on the region of interest of the multidirectional characteristic diagram to obtain a binary image, carrying out nearest neighbor clustering on data points with the gray value of 255 in the binary image, and extracting the weld seam contour according to the obtained coordinates of the matching position and the clustering result.
Further, the automatic determination of the filtering angle and the region of interest of the Gabor filtering by using the reference image comprises the following steps:
s11, the welding gun is in the initial welding position, the laser vision sensor is in the working state, and a welding seam image without arc light and with laser stripes is collected without arc striking and is used as a reference image for extracting the contour of the welding seam at this time;
s12, performing self-adaptive local threshold segmentation processing on the reference image to obtain a binary strip-shaped weld seam outline;
s13, determining 80 pixel boundaries respectively from top to bottom by taking the center line of the strip-shaped welding seam contour as a reference, and taking the boundaries as the top and bottom boundaries of the region of interest; the left and right boundaries of the region of interest are the left and right boundaries of the reference center line;
s14, calculating the slope of the data where the central line is located:
wherein
A footer to represent the data is indicated,
indicating foot markThe data points of the surrounding neighbors are,
coordinates representing the row on which the laser stripe data point lies,
coordinates representing a column in which the laser stripe data point is located;
s15, obtaining monotonously increasing and decreasing slope intervals by using formulas (2) and (3), respectively:
（2）
where t represents the angle of slope;
s16, obtaining the length of the slope monotone interval by adopting the following method: subtracting the subscript of the first slope from the subscript of the last slope of each slope monotonic interval;
s17, calculating the average length of all slope monotonous intervals, and acquiring the part of the slope monotonous interval with the length larger than the average length;
s18, carrying out segmentation by an Otsu threshold segmentation method on the numerical value of the slope monotonic interval with the length larger than the average length to obtain a threshold, wherein the interval with the length numerical value larger than the threshold represents the characteristic point of the weld contour at the position, and the position of the midpoint of the interval corresponding to the position in the image is taken as the position of the characteristic point;
s19, dividing the weld contour central line extracted from the reference image into a plurality of sections by the identified characteristic points, and determining the arctangent angle of the slope of each section by the coordinates of the first data point and the last data point of each section from left to right as follows:
wherein
Indicates the number of segments of the above-mentioned segment,
andrespectively the ordinate and abscissa of the first data point of each segment,and
respectively, the ordinate and abscissa of the last data point of the segment.
Further, the step S2 includes:
acquiring a local threshold:
whereinAn image of the weld is represented,
which represents the number of lines of the image,
indicating the position of the line on which the image is looped each time,
which represents the number of columns of the image,
indicating the area covered by each circular column,
indicating that the interval of the columns is 2 per cycle,
the number of rows representing the weld image,
the number of columns representing the weld image.
Further, the acquiring of the reference image and the region of interest of the directional feature map specifically includes:
s21, using the above
Respectively carrying out Gabor filtering on the welding seam images collected in welding, and obtaining a direction characteristic diagram with the same size as the original image every time, wherein the Gabor filtering formula is as follows:
wherein the content of the first and second substances,
and
is the coordinate of each pixel point in the collected welding seam image,
is the frequency of the filtering of the signal,
is said filteringThe angle of the angle is set to be,
is the standard deviation, and set
；
S22, dividing the welding seam image into a plurality of welding seam images by taking the vertical coordinate of the obtained characteristic points as a boundary line
Sequentially intercepting the direction characteristic diagrams of all the areas to form a multidirectional characteristic diagram;
and S23, respectively intercepting the multi-direction feature map and the reference image according to the region of interest to obtain the regions of interest of the direction feature map and the reference image.
Further, the extracting the weld seam profile according to the obtained coordinates of the matching position and the clustering result specifically comprises:
s31, performing self-adaptive local threshold segmentation processing on the region of interest of the multidirectional feature map to obtain a binary image, setting Euclidean distance of 5 pixels as a clustering threshold, and performing nearest neighbor clustering on data points with the gray value of 255 to generate a certain number of classes;
s32, in the scale invariant feature transformation algorithm, the coordinates of the matching positions generated in the two input interested areas are superposed on the binary image;
and S33, extracting the class which is closest to the coordinates of each matching position and is less than 5 pixels away, wherein all the extracted classes form the weld seam outline.
The invention has the advantages that:
the invention relates to a general method for extracting a weld contour of GMAW (gas metal arc welding) automatic welding of a typical joint of a thick plate under laser vision sensing, which is characterized in that a weld image that a welding gun is at an initial welding position and a laser vision sensing system is not arcing under a working state is collected and used as a reference image, adaptive local threshold segmentation processing is carried out on the image to obtain the weld contour, and an interested area is determined according to the weld contour; filtering a weld image acquired in real time in welding by adopting a Gabor filter with a self-adaptive filtering angle to generate a multidirectional characteristic diagram; respectively acquiring a reference image F1 and a multidirectional feature map F2 in the region by using the acquired region of interest; processing the F1 and the F2 by using a scale invariant feature transform algorithm to obtain the coordinates (xi, yi) of the similar point of the two images (i =1,2, …); performing self-adaptive local threshold segmentation processing on the F2 to obtain a binary weld image; and (3) carrying out nearest neighbor clustering on data points in the binary image, and extracting classes belonging to the weld seam contour according to coordinates (xi, yi) (i =1,2, …), wherein the classes form the whole weld seam contour. The invention is effective to typical butt joint and angle joint joints, can be popularized to the extraction of welding seams of other joint outlines, is beneficial to reducing the software development cost in welding and is beneficial to the integration of a welding information acquisition and processing system.
In addition to the objects, features and advantages described above, other objects, features and advantages of the present invention are also provided. The present invention will be described in further detail below with reference to the drawings.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this application, illustrate embodiments of the invention and, together with the description, serve to explain the invention and not to limit the invention.
FIG. 1 is a flow chart of a general method of automatic weld profile extraction of the present invention;
FIG. 2 is a diagram of the information provided by the T-junction reference image of the present invention;
FIG. 3 is a schematic representation of the multi-directional multi-zone profile determination of the present invention;
FIG. 4 is a reference image and feature map region of interest transformed based on scale invariant features of the present invention
A matching graph of domain affinity locations;
FIG. 5 is an extracted graph of weld profiles based on matching locations and nearest neighbor clustering in accordance with the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
Referring to fig. 1, as shown in fig. 1, a general method for automatic weld profile extraction includes the following steps:
collecting a welding seam image which is not subjected to arc striking before welding as a reference image, and automatically determining a filtering angle and an interested area of Gabor filtering by using the reference image; collecting a welding image, carrying out self-adaptive Gabor filtering on the welding image to obtain a multi-direction characteristic diagram, and simultaneously obtaining a reference image and an interested area part of the direction characteristic diagram; processing the two interested area parts by using a scale invariant feature transform algorithm to obtain the matching positions of the similar points of the two images; and (3) carrying out self-adaptive local threshold segmentation processing on the region of interest of the multidirectional characteristic diagram to obtain a binary image, carrying out nearest neighbor clustering on data points with the gray value of 255 in the binary image, and extracting the weld seam contour according to the obtained coordinates of the matching position and the clustering result.
The method for automatically determining the filtering angle and the region of interest of the Gabor filtering by utilizing the reference image comprises the following steps:
s11, the welding gun is in the initial welding position, the laser vision sensor is in the working state, and a welding seam image without arc light and with laser stripes is collected without arc striking and is used as a reference image for extracting the contour of the welding seam at this time;
s12, performing self-adaptive local threshold segmentation processing on the reference image to obtain a binary strip-shaped weld seam outline;
s13, determining 80 pixel boundaries respectively from top to bottom by taking the center line of the strip-shaped welding seam contour as a reference, and taking the boundaries as the top and bottom boundaries of the region of interest; the left and right boundaries of the region of interest are the left and right boundaries of the reference center line;
s14, calculating the slope of the data where the central line is located:
wherein
A footer to represent the data is indicated,
indicating foot mark
The data points of the surrounding neighbors are,
coordinates representing the row on which the laser stripe data point lies,
coordinates representing a column in which the laser stripe data point is located;
s15, obtaining monotonously increasing and decreasing slope intervals by using formulas (2) and (3), respectively:
where t represents the angle of slope;
s16, obtaining the length of the slope monotone interval by adopting the following method: subtracting the subscript of the first slope from the subscript of the last slope of each slope monotonic interval;
s17, calculating the average length of all slope monotonous intervals, and acquiring the part of the slope monotonous interval with the length larger than the average length;
s18, carrying out segmentation by an Otsu threshold segmentation method on the numerical value of the slope monotonic interval with the length larger than the average length to obtain a threshold, wherein the interval with the length numerical value larger than the threshold represents the characteristic point of the weld contour at the position, and the position of the midpoint of the interval corresponding to the position in the image is taken as the position of the characteristic point;
s19, dividing the weld contour central line extracted from the reference image into a plurality of sections by the identified characteristic points, and determining the arctangent angle of the slope of each section by the coordinates of the first data point and the last data point of each section from left to right as follows:
wherein
Indicates the number of segments of the above-mentioned segment,and
respectively the ordinate and abscissa of the first data point of each segment,and
respectively, the ordinate and abscissa of the last data point of the segment.
The step S2 includes:
acquiring a local threshold:
wherein
An image of the weld is represented,
which represents the number of lines of the image,
indicating the position of the line on which the image is looped each time,
which represents the number of columns of the image,
indicating the area covered by each circular column,
indicating that the interval of the columns is 2 per cycle,
the number of rows representing the weld image,
the number of columns representing the weld image.
The obtaining of the reference image and the region of interest of the direction feature map specifically includes:
s21, using the above
Respectively carrying out Gabor filtering on the welding seam images collected in welding, and obtaining a direction characteristic diagram with the same size as the original image every time, wherein the Gabor filtering formula is as follows:
wherein the content of the first and second substances,
and
is the coordinate of each pixel point in the collected welding seam image,
is the frequency of the filtering of the signal,
is the angle of the filtering, and is,
is the standard deviation, and set
；
S22, dividing the welding seam image into a plurality of welding seam images by taking the vertical coordinate of the obtained characteristic points as a boundary line
Sequentially intercepting the direction characteristic diagrams of all the areas to form a multidirectional characteristic diagram;
and S23, respectively intercepting the multi-direction feature map and the reference image according to the region of interest to obtain the regions of interest of the direction feature map and the reference image.
The extracting of the weld seam profile according to the obtained coordinates of the matching position and the clustering result specifically comprises:
s31, performing self-adaptive local threshold segmentation processing on the region of interest of the multidirectional feature map to obtain a binary image, setting Euclidean distance of 5 pixels as a clustering threshold, and performing nearest neighbor clustering on data points with the gray value of 255 to generate a certain number of classes;
s32, in the scale invariant feature transformation algorithm, the coordinates of the matching positions generated in the two input interested areas are superposed on the binary image;
and S33, extracting the class which is closest to the coordinates of each matching position and is less than 5 pixels away, wherein all the extracted classes form the weld seam outline.
The invention provides a general method for extracting a weld contour based on laser vision sensing, which is based on typical butt joint and fillet joint weld contour of thick plate welding. The method fully utilizes the non-arc welding seam outline image collected before each welding, and combines a direction detection algorithm and a scale invariant feature transformation algorithm to position the welding seam outline in the welding. The specific technical scheme is as follows:
step 1: before each welding, the laser vision sensor is in a working state, the welding gun is in a welding state, and an outline image of a welding seam which is not subjected to arcing is collected and used as a reference image;
step 2: adopting self-adaptive local threshold segmentation processing to the reference image to obtain a banded weld outline in the binary image, and obtaining a central line of the banded weld outline by calculating an average coordinate value of each row of banded weld outlines, wherein a calculation method for determining a threshold in the self-adaptive local threshold segmentation processing is a formula (1);
wherein
An image of the weld is represented,
which represents the number of lines of the image,
indicating the position of the line on which the image is looped each time,
which represents the number of columns of the image,
indicating the area covered by each circular column,
indicating that the interval of the columns is 2 per cycle,
the number of rows representing the weld image,
the number of columns representing the weld image.
And step 3: determining an interested area by taking the central line as a reference and 80 pixels above and below the central line, extracting a part of an original reference image in the interested area for subsequent matching processing, and calling the part to be F1;
and 4, step 4: obtaining the slope of the central line of the weld seam outline by using a formula (2):
whereinA footer to represent the data is indicated,
indicating foot mark
The data points of the surrounding neighbors are,
coordinates representing the row on which the laser stripe data point lies,
coordinates representing a column in which the laser stripe data point is located;
and 5: obtaining monotone increasing and decreasing slope intervals by respectively adopting formulas (3) and (4):
step 6: obtaining length values of slope intervals which are monotonically increased and decreased, processing the length values by utilizing an Otsu threshold segmentation method to obtain a threshold of the length value, and determining the interval with the length value larger than the threshold, which is called as a large interval;
and 7: determining the position of the middle position of the large region in the image, wherein the position is the position of the feature point;
and 8: dividing the central line of the welding seam contour in the reference image into m segments by the characteristic points, and determining the arctangent values of the slopes corresponding to the m segments by using a formula (5), wherein the values are the filtering angles of Gabor filtering:
and step 9: carrying out m times of Gabor filtering on the collected images in welding by using formulas (6) and (7) to obtain m different direction characteristic graphs, wherein the filtering angle at each time is
：
Wherein x and y are coordinates of elements in the grayscale image,
is the frequency of the filtering of the signal,
is the angle of the filtering, and,
is a standard deviation, and
。
step 10: according to the segmentation in the step 8, sequentially intercepting the corresponding direction feature maps in the step 9 from left to right to form a final multidirectional feature map;
step 11: intercepting a multi-directional feature map portion in the region of interest, referred to as "F2";
step 12: performing self-adaptive local threshold segmentation processing on the F2 to obtain a binary image;
step 13: setting Euclidean distance as 5 pixel values as a threshold value in clustering, and performing nearest neighbor clustering on data points with the gray value of 255 in the binary image in the step 12 to obtain different classes;
step 14: processing the F1 in the step 3 and the F2 in the step 11 by adopting a scale invariant feature transform algorithm to obtain the positions of matching points at which the two images are similar;
step 15: superposing the coordinates of the matching points obtained in the two images on the binary image in the step 12;
step 16: and (3) calculating the distance between each class in the step (13) and all the matching points in the step (14), and if the distance between a certain class and one of the matching points is less than 5 pixels, extracting the class as a part of the weld contour, and accordingly extracting all the classes, wherein the classes form the final weld contour.
Example 1:
1. the welding gun is in a welding posture and is positioned at an initial welding position, the laser vision sensing system is in a working state, and an image of a welding seam which is not subjected to arcing before welding is collected and used as a reference image, as shown in the leftmost image of the image in FIG. 2;
2. the reference image is processed as shown in fig. 2, that is, adaptive local threshold segmentation processing and strip weld contour center acquisition processing are respectively performed, and then an area of interest, a feature point and a filtering angle are respectively acquired
（m=1,2,…）；
3. Using the obtained filter angle
(m =1,2, …) carrying out Gabor filtering, processing a welding seam image acquired in the welding process, and determining a final multidirectional feature map by combining the region of interest and the region segmented by the feature points, as shown in FIG. 3;
4. taking the interesting region part of the reference image and the multidirectional feature map as input, and processing by using a scale-invariant feature transformation algorithm to obtain the coordinates of similar matching points, as shown in FIG. 4;
5. performing self-adaptive local threshold segmentation processing on the multidirectional feature map, performing nearest neighbor clustering on data points with a segmented gray value of 255, wherein Euclidean distance threshold for clustering is 5 pixels, and circles with different radiuses in the graph 5 represent classes in a clustering result;
6. according to the nearest principle, classes which are nearest to the matching point in Euclidean distance and within 5 pixels are obtained, and the classes are weld contour data, and are shown in a right diagram in FIG. 5.
The invention relates to a general method for extracting a weld contour of GMAW (gas metal arc welding) automatic welding of a typical joint of a thick plate under laser vision sensing, which is characterized in that a weld image that a welding gun is at an initial welding position and a laser vision sensing system is not arcing under a working state is collected and used as a reference image, adaptive local threshold segmentation processing is carried out on the image to obtain the weld contour, and an interested area is determined according to the weld contour; filtering a weld image acquired in real time in welding by adopting a Gabor filter with a self-adaptive filtering angle to generate a multidirectional characteristic diagram; respectively acquiring a reference image F1 and a multidirectional feature map F2 in the region by using the acquired region of interest; processing the F1 and the F2 by using a scale invariant feature transform algorithm to obtain the coordinates (xi, yi) of the similar point of the two images (i =1,2, …); performing self-adaptive local threshold segmentation processing on the F2 to obtain a binary weld image; and (3) carrying out nearest neighbor clustering on data points in the binary image, and extracting classes belonging to the weld seam contour according to coordinates (xi, yi) (i =1,2, …), wherein the classes form the whole weld seam contour. The invention is effective to typical butt joint and angle joint joints, can be popularized to the extraction of welding seams of other joint outlines, is beneficial to reducing the software development cost in welding and is beneficial to the integration of a welding information acquisition and processing system.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.
Claims (5)
1. A general method for automatically extracting welding seam outline is characterized by comprising the following steps
The method comprises the following steps:
collecting a welding seam image which is not subjected to arc striking before welding as a reference image, and automatically determining a filtering angle and an interested area of Gabor filtering by using the reference image; collecting a welding image, carrying out self-adaptive Gabor filtering on the welding image to obtain a multi-direction characteristic diagram, and simultaneously obtaining a reference image and an interested area part of the direction characteristic diagram; processing the two interested area parts by using a scale invariant feature transform algorithm to obtain the matching positions of the similar points of the two images; and (3) carrying out self-adaptive local threshold segmentation processing on the region of interest of the multidirectional characteristic diagram to obtain a binary image, carrying out nearest neighbor clustering on data points with the gray value of 255 in the binary image, and extracting the weld seam contour according to the obtained coordinates of the matching position and the clustering result.
2. The general method for automatic weld profile extraction according to claim 1, characterized in that
In that, the automatic determination of the filtering angle and the region of interest of the Gabor filtering by using the reference image comprises the following steps:
s11, the welding gun is in the initial welding position, the laser vision sensor is in the working state, and a welding seam image without arc light and with laser stripes is collected without arc striking and is used as a reference image for extracting the contour of the welding seam at this time;
s12, performing self-adaptive local threshold segmentation processing on the reference image to obtain a binary strip-shaped weld seam outline;
s13, determining 80 pixel boundaries respectively from top to bottom by taking the center line of the strip-shaped welding seam contour as a reference, and taking the boundaries as the top and bottom boundaries of the region of interest; the left and right boundaries of the region of interest are the left and right boundaries of the reference center line;
s14, calculating the slope of the data where the central line is located:
（1）
wherein
A footer to represent the data is indicated,
indicating foot mark
The data points of the surrounding neighbors are,
coordinates representing the row on which the laser stripe data point lies,
coordinates representing a column in which the laser stripe data point is located;
s15, obtaining monotonously increasing and decreasing slope intervals by using formulas (2) and (3), respectively:
where t represents the angle of slope;
s16, obtaining the length of the slope monotone interval by adopting the following method: subtracting the subscript of the first slope from the subscript of the last slope of each slope monotonic interval;
s17, calculating the average length of all slope monotonous intervals, and acquiring the part of the slope monotonous interval with the length larger than the average length;
s18, carrying out segmentation by an Otsu threshold segmentation method on the numerical value of the slope monotonic interval with the length larger than the average length to obtain a threshold, wherein the interval with the length numerical value larger than the threshold represents the characteristic point of the weld contour at the position, and the position of the midpoint of the interval corresponding to the position in the image is taken as the position of the characteristic point;
s19, dividing the weld contour central line extracted from the reference image into a plurality of sections by the identified characteristic points, and determining the arctangent angle of the slope of each section by the coordinates of the first data point and the last data point of each section from left to right as follows:
whereinIndicates the number of segments of the above-mentioned segment,
and
respectively the ordinate and abscissa of the first data point of each segment,
and
respectively, the ordinate and abscissa of the last data point of the segment.
3. The general method for automatic weld profile extraction according to claim 2, characterized in that
In that, the step S2 includes:
acquiring a local threshold:
wherein
An image of the weld is represented,which represents the number of lines of the image,
indicating the position of the line on which the image is looped each time,
which represents the number of columns of the image,
indicating the area covered by each circular column,
indicating that the interval of the columns is 2 per cycle,
the number of rows representing the weld image,
the number of columns representing the weld image.
4. The general method for automatic weld profile extraction according to claim 1 or 2, which is based on the following claims
The method is characterized in that the obtaining of the reference image and the region of interest of the direction feature map specifically comprises the following steps:
s21, using the above
Respectively carrying out Gabor filtering on the welding seam images collected in welding, and obtaining a direction characteristic diagram with the same size as the original image every time, wherein the Gabor filtering formula is as follows:
wherein the content of the first and second substances,
and
is the coordinate of each pixel point in the collected welding seam image,
is the frequency of the filtering of the signal,is the angle of the filtering, and is,
is the standard deviation, and set
；
S22, dividing the welding seam image into a plurality of welding seam images by taking the vertical coordinate of the obtained characteristic points as a boundary line
Sequentially intercepting the direction characteristic diagrams of all the areas to form a multidirectional characteristic diagram;
and S23, respectively intercepting the multi-direction feature map and the reference image according to the region of interest to obtain the regions of interest of the direction feature map and the reference image.
5. The general method for automatic weld profile extraction according to claim 1, characterized in that
The step of extracting the weld contour according to the obtained coordinates of the matching position and the clustering result specifically comprises the following steps:
s31, performing self-adaptive local threshold segmentation processing on the region of interest of the multidirectional feature map to obtain a binary image, setting Euclidean distance of 5 pixels as a clustering threshold, and performing nearest neighbor clustering on data points with the gray value of 255 to generate a certain number of classes;
s32, in the scale invariant feature transformation algorithm, the coordinates of the matching positions generated in the two input interested areas are superposed on the binary image;
and S33, extracting the class which is closest to the coordinates of each matching position and is less than 5 pixels away, wherein all the extracted classes form the weld seam outline.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911194751.0A CN110814465B (en) | 2019-11-28 | 2019-11-28 | Universal method for automatically extracting welding seam contour |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911194751.0A CN110814465B (en) | 2019-11-28 | 2019-11-28 | Universal method for automatically extracting welding seam contour |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110814465A true CN110814465A (en) | 2020-02-21 |
| CN110814465B CN110814465B (en) | 2021-09-03 |
Family
ID=69543023
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911194751.0A Active CN110814465B (en) | 2019-11-28 | 2019-11-28 | Universal method for automatically extracting welding seam contour |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110814465B (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111414912A (en) * | 2020-03-31 | 2020-07-14 | 北京博清科技有限公司 | Method, device and equipment for identifying characteristic points of butt-joint type welding seam and storage medium |
| CN111738991A (en) * | 2020-06-04 | 2020-10-02 | 西安数合信息科技有限公司 | Method for creating digital ray detection model of weld defects |
| CN112222569A (en) * | 2020-09-30 | 2021-01-15 | 北京博清科技有限公司 | Welding seam tracking method and device, robot and storage medium |
| CN112365439A (en) * | 2020-09-24 | 2021-02-12 | 南昌大学 | Method for synchronously detecting forming characteristics of GMAW (gas metal arc welding) weld of galvanized steel and welding gun direction in real time |
| CN112837307A (en) * | 2021-02-24 | 2021-05-25 | 北京博清科技有限公司 | Method, device, processor and system for determining welding bead profile |
| CN113436209A (en) * | 2021-06-23 | 2021-09-24 | 南通大学 | Novel welding line center line extraction method based on layer-by-layer indentation strategy |
| CN116452586A (en) * | 2023-06-15 | 2023-07-18 | 山东飞宏工程机械有限公司 | Automatic butt welding quality detection system for tunnel small guide pipe residual materials |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103008881A (en) * | 2012-12-05 | 2013-04-03 | 中国电子科技集团公司第四十五研究所 | Seam tracking method based on template matching |
| CN104084669A (en) * | 2014-06-20 | 2014-10-08 | 河北工业大学 | Visual detecting method and device based on decoupling detection for circular seams |
| CN106952281A (en) * | 2017-05-15 | 2017-07-14 | 上海交通大学 | A kind of method that weld profile feature recognition and its welding bead are planned in real time |
| CN107824940A (en) * | 2017-12-07 | 2018-03-23 | 淮安信息职业技术学院 | Welding seam traking system and method based on laser structure light |
| CN107914067A (en) * | 2017-10-31 | 2018-04-17 | 南昌大学 | A kind of welding gun deviation three-dimensional extracting method of the plate sheet welding based on passive vision sensing |
| DE102017203886A1 (en) * | 2017-03-09 | 2018-09-13 | Bayerische Motoren Werke Aktiengesellschaft | Joining method for materially joining at least two components and component assembly |
| CN109615603A (en) * | 2018-10-27 | 2019-04-12 | 南昌大学 | A kind of visual attention model of task based access control driving extracts the universal method of laser stripe |
| CN109604777A (en) * | 2017-12-07 | 2019-04-12 | 淮安信息职业技术学院 | Welding seam traking system and method based on laser structure light |
| CN109986172A (en) * | 2019-05-21 | 2019-07-09 | 广东工业大学 | A kind of weld and HAZ method, equipment and system |
-
2019
- 2019-11-28 CN CN201911194751.0A patent/CN110814465B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103008881A (en) * | 2012-12-05 | 2013-04-03 | 中国电子科技集团公司第四十五研究所 | Seam tracking method based on template matching |
| CN104084669A (en) * | 2014-06-20 | 2014-10-08 | 河北工业大学 | Visual detecting method and device based on decoupling detection for circular seams |
| DE102017203886A1 (en) * | 2017-03-09 | 2018-09-13 | Bayerische Motoren Werke Aktiengesellschaft | Joining method for materially joining at least two components and component assembly |
| CN106952281A (en) * | 2017-05-15 | 2017-07-14 | 上海交通大学 | A kind of method that weld profile feature recognition and its welding bead are planned in real time |
| CN107914067A (en) * | 2017-10-31 | 2018-04-17 | 南昌大学 | A kind of welding gun deviation three-dimensional extracting method of the plate sheet welding based on passive vision sensing |
| CN107824940A (en) * | 2017-12-07 | 2018-03-23 | 淮安信息职业技术学院 | Welding seam traking system and method based on laser structure light |
| CN109604777A (en) * | 2017-12-07 | 2019-04-12 | 淮安信息职业技术学院 | Welding seam traking system and method based on laser structure light |
| CN109615603A (en) * | 2018-10-27 | 2019-04-12 | 南昌大学 | A kind of visual attention model of task based access control driving extracts the universal method of laser stripe |
| CN109986172A (en) * | 2019-05-21 | 2019-07-09 | 广东工业大学 | A kind of weld and HAZ method, equipment and system |
Non-Patent Citations (1)
| Title |
|---|
| 陈海永等: "基于焊缝兴趣点的多尺度形状描述符模板匹配算法", 《焊接学报》 * |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111414912A (en) * | 2020-03-31 | 2020-07-14 | 北京博清科技有限公司 | Method, device and equipment for identifying characteristic points of butt-joint type welding seam and storage medium |
| CN111738991A (en) * | 2020-06-04 | 2020-10-02 | 西安数合信息科技有限公司 | Method for creating digital ray detection model of weld defects |
| CN112365439A (en) * | 2020-09-24 | 2021-02-12 | 南昌大学 | Method for synchronously detecting forming characteristics of GMAW (gas metal arc welding) weld of galvanized steel and welding gun direction in real time |
| CN112365439B (en) * | 2020-09-24 | 2023-10-03 | 南昌大学 | Method for synchronously detecting forming characteristics of GMAW welding seam of galvanized steel and direction of welding gun in real time |
| CN112222569A (en) * | 2020-09-30 | 2021-01-15 | 北京博清科技有限公司 | Welding seam tracking method and device, robot and storage medium |
| CN112837307A (en) * | 2021-02-24 | 2021-05-25 | 北京博清科技有限公司 | Method, device, processor and system for determining welding bead profile |
| CN113436209A (en) * | 2021-06-23 | 2021-09-24 | 南通大学 | Novel welding line center line extraction method based on layer-by-layer indentation strategy |
| CN113436209B (en) * | 2021-06-23 | 2023-11-17 | 南通大学 | Novel weld joint center line extraction method based on layer-by-layer indentation strategy |
| CN116452586A (en) * | 2023-06-15 | 2023-07-18 | 山东飞宏工程机械有限公司 | Automatic butt welding quality detection system for tunnel small guide pipe residual materials |
| CN116452586B (en) * | 2023-06-15 | 2023-09-26 | 山东飞宏工程机械有限公司 | Automatic butt welding quality detection system for tunnel small guide pipe residual materials |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110814465B (en) | 2021-09-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110814465B (en) | Universal method for automatically extracting welding seam contour | |
| CN101887586B (en) | Self-adaptive angular-point detection method based on image contour sharpness | |
| US8401333B2 (en) | Image processing method and apparatus for multi-resolution feature based image registration | |
| Yuan et al. | Robust lane detection for complicated road environment based on normal map | |
| CN108898147B (en) | Two-dimensional image edge flattening method and device based on angular point detection | |
| CN110717872B (en) | Method and system for extracting characteristic points of V-shaped welding seam image under laser-assisted positioning | |
| CN110610150B (en) | Tracking method, device, computing equipment and medium of target moving object | |
| JPH08327296A (en) | Method and equipment for detecting target motion and applying method of said method | |
| CN113706581B (en) | Target tracking method based on residual channel attention and multi-level classification regression | |
| CN107230219B (en) | Target person finding and following method on monocular robot | |
| CN111667470B (en) | Industrial pipeline flaw detection inner wall detection method based on digital image | |
| CN111753693B (en) | Target detection method under static scene | |
| CN108196729A (en) | A kind of finger tip point rapid detection method based on infrared video | |
| CN113012181B (en) | Novel quasi-circular detection method based on Hough transformation | |
| CN113989604A (en) | Tire DOT information identification method based on end-to-end deep learning | |
| JP3659426B2 (en) | Edge detection method and edge detection apparatus | |
| CN110717910B (en) | CT image target detection method based on convolutional neural network and CT scanner | |
| CN109615603B (en) | Method for extracting laser stripes based on task-driven visual attention model | |
| Jiang et al. | Mobile robot gas source localization via top-down visual attention mechanism and shape analysis | |
| Truong et al. | Lane boundaries detection algorithm using vector lane concept | |
| CN114529715B (en) | Image identification method and system based on edge extraction | |
| CN105930813B (en) | A method of detection composes a piece of writing this under any natural scene | |
| CN113836978A (en) | Road area determination device and method and electronic equipment | |
| EP3745300A1 (en) | Obstacle recognition method for library robot based on depth image | |
| CN111209872A (en) | Real-time rolling fingerprint splicing method based on dynamic programming and multi-objective optimization |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |