CN113762400A - Weld joint position autonomous extraction method based on naive Bayes classifier - Google Patents
Weld joint position autonomous extraction method based on naive Bayes classifier Download PDFInfo
- Publication number
- CN113762400A CN113762400A CN202111066499.2A CN202111066499A CN113762400A CN 113762400 A CN113762400 A CN 113762400A CN 202111066499 A CN202111066499 A CN 202111066499A CN 113762400 A CN113762400 A CN 113762400A
- Authority
- CN
- China
- Prior art keywords
- data
- welding seam
- weld
- class
- interference
- 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
- 238000000605 extraction Methods 0.000 title claims abstract description 32
- 238000003466 welding Methods 0.000 claims abstract description 60
- 238000000034 method Methods 0.000 claims abstract description 21
- 230000011218 segmentation Effects 0.000 claims abstract description 21
- 230000000007 visual effect Effects 0.000 claims abstract description 15
- 238000010891 electric arc Methods 0.000 claims abstract description 9
- 238000010586 diagram Methods 0.000 claims abstract description 8
- 238000001914 filtration Methods 0.000 claims abstract description 7
- 210000001503 joint Anatomy 0.000 claims description 6
- 239000013598 vector Substances 0.000 claims description 6
- 238000004422 calculation algorithm Methods 0.000 claims description 5
- 238000001514 detection method Methods 0.000 claims description 5
- 238000007635 classification algorithm Methods 0.000 claims description 4
- 241000764238 Isis Species 0.000 claims description 2
- 238000004364 calculation method Methods 0.000 claims description 2
- 230000008859 change Effects 0.000 claims description 2
- 238000010276 construction Methods 0.000 claims description 2
- 238000009499 grossing Methods 0.000 claims description 2
- 230000035772 mutation Effects 0.000 claims description 2
- 230000000630 rising effect Effects 0.000 claims description 2
- 230000008569 process Effects 0.000 abstract description 6
- 230000002411 adverse Effects 0.000 abstract description 2
- 230000000694 effects Effects 0.000 abstract description 2
- 238000012545 processing Methods 0.000 description 5
- 230000002401 inhibitory effect Effects 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 239000011324 bead Substances 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 239000002184 metal Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000000877 morphologic effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
- G06F18/24155—Bayesian classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Abstract
The invention provides an autonomous extraction method of a welding seam position based on a naive Bayes classifier, which comprises the steps of firstly designing a plurality of improved Gabor filters which are more effective in distinguishing direction characteristics, and linearly combining filtering results to generate a comprehensive direction characteristic diagram of the welding seam position; then, carrying out local threshold autonomous segmentation on the data, designing probability density functions for distinguishing targets and interferences from the thickness, uniformity and compactness of the data respectively through a nearest neighbor clustering and supervision method, and preliminarily realizing effective judgment of the positions of welding seams and the interferences based on a naive Bayes classifier; and finally, expanding visual feature competition for each data class overlapped in the horizontal direction, and further removing interference. The invention provides a typical joint weld position extraction method, which can effectively overcome adverse effects caused by interference such as electric arc, splashing and the like, can stabilize weld tracking in an automatic electric arc welding process based on laser vision sensing, and improve welding quality.
Description
Technical Field
The invention relates to the technical field of automatic welding, in particular to an autonomous extraction method of a welding seam position based on a naive Bayes classifier.
Background
The visual sensing in the current automatic arc welding is still one of the main information acquisition means, and the laser visual sensing in the thick plate automatic arc welding is considered to be one of the effective methods for effectively detecting the position to be welded and the weld forming characteristics. In this process, a laser beam is placed in front of the welding torch and the area to be welded is detected perpendicular to the welding direction. The extraction of the weld seam position (laser profile) is therefore a prerequisite for the subsequent on-line intervention of the welding process. However, uncertainty of the welding process, strong spatter, and the variable nature of the thick plate weld profile make extracting the effective weld location in real time still challenging.
In order to suppress high-frequency noise in a weld image in weld position extraction, various filters are applied to various weld position extraction schemes to remove significant noise. For the welding seam image obtained by adopting the arc light shielding measure, the welding seam outline can be obtained by adopting a traversal maximum gray value searching method. For more noisy images, thresholding is typically used to further eliminate the interference and reduce the data processing dimension. And finally, extracting the position of the welding line by adopting a morphological extraction method, a model matching reconstruction method, a clustering indirect extraction method and the like.
The groove of the weld contour aimed at by the existing method is small (the plate thickness is less than 40mm), the space span of the weld contour is small, and the appearance of the weld is basically unchanged in welding. In addition, because the arc is shielded, the weld images have less interference. In the thick plate automatic welding based on laser visual sensing, the laser detection span is large, the interference is more complicated due to the large welding seam outline, the splashing is more prone to contaminating the welding seam outline, and the appearance and the position of a multi-welding seam are changed along with the number of welding passes. The above situation makes effective extraction of the weld location more challenging.
Disclosure of Invention
The invention provides an autonomous extraction method of a weld position based on a Naive Bayes Classifier (Naive Bayes Classifier), aiming at the problems that in thick plate gas metal arc gas protection automatic welding based on laser visual sensing, the weld span is large, and complete arc and splash interference and changeable weld outlines exist in a weld image. The method comprises the steps of firstly, designing a plurality of improved Gabor filters to effectively detect the direction characteristics of a weld contour, preliminarily inhibiting electric arc and splash interference, then designing a probability density function for effectively distinguishing three visual characteristics of laser stripes and interference data in a supervision mode, further inhibiting the interference on two-dimensional data by using a naive Bayes classifier, and finally finishing final interference removal according to the spatial span of a data cluster and the contour fluctuation degree of the data cluster. The effectiveness and the anti-interference capability of the algorithm provided by the invention are verified by the extraction test of the positions of the welding seams with different shapes of the T-shaped thick plate, the butt joint and the lap joint of the thin plate.
In order to achieve the purpose, the invention provides the following technical scheme: an autonomous extraction method of a weld joint position based on a naive Bayes classifier comprises the following steps:
designing an improved Gabor filter to initially inhibit strong electric arc and splash interference of a welding seam image;
step two, performing local threshold autonomous segmentation on the filtered image, and realizing data classification according to a nearest neighbor clustering algorithm; three prior probability density functions of thickness, uniformity and compactness which are subject to normal distribution and are used for distinguishing target data and interference data are designed according to data classification, and preliminary identification of the welding seam outline is realized on the basis of a naive Bayes classification algorithm according to a maximum posterior probability criterion;
and step three, designing a visual feature calculation method of space span and fluctuation degree, performing overlapped unfolding visual feature competition on the preliminarily identified data in the horizontal direction, and finally accurately extracting the position of the welding seam.
Further, an improved Gabor filter is designed in the first step, strong arc and splash interference of a weld image is preliminarily inhibited, and the specific steps are as follows:
the first step is as follows: the improved construction method of the convolution kernel of the two-dimensional Gabor filter comprises the following steps:
where x 'is x cos θ + y sin θ, y' is-xsin θ + ycos θ, f is the frequency in pixels, θ is the detection direction, σ is the convolution template size,
is a phase, designed specifically as
σ＝4，
n and m are designed as follows:
the contrast ratio of the set direction area and the background can be obviously increased by the constructed convolution kernel, and the direction characteristic of the welding seam position can be more effectively highlighted;
the second step is that: for the extraction of the welding seam position of the T-shaped joint, formulas (2) a-c are respectively used for detecting a web plate, a groove and a welding seam area, and formula (2) b is also used for detecting a bottom plate;set the filter angles to be theta1∈[-20°,-5°]、θ2∈[-80°,-110°]、θ3∈[5°,15°]And theta4∈[15°,35°](ii) a For the extraction of the position of the V-shaped groove weld of the butt joint, formulas (3) a-c are respectively used for detecting a plate surface and left and right groove areas;
the third step: and carrying out linear combination on the filtering results to generate a comprehensive direction characteristic diagram.
Further, in the second step, the filtered image is subjected to local threshold autonomous segmentation, and data classification is realized according to a nearest neighbor clustering algorithm, and the specific steps are as follows:
the first step is as follows: performing local threshold autonomous segmentation on the comprehensive direction characteristic graph to realize data simplification, wherein the specific local threshold determination method comprises the following steps:
adopting 5 with step length of 5╳5, smoothing the image F in the vertical direction by the window to obtain average gray values at different height positions
Wherein I represents a row, j represents a column, and the vector I formed by average gray values obtained at each position is subjected to linear filtering twice, and the length of each filter is more than 19;
obtaining each monotone increasing interval of the filtered vector I', recording the starting position and the ending position of the monotone increasing interval, and simultaneously recording the gray value of the corresponding position;
calculating the difference value of the gray value at the two ends of each monotonous interval and recording the monotonous interval D with the difference value larger than G pixelsiThereby representing the gray value mutation area;
fourthly, calculating the monotonous interval DiThe difference value of the gray values of the vectors I and I' at each termination position is recorded again, and the monotone interval D with the difference value larger than G pixels is recorded againi', the abrupt change of the original gray value rising region;
using each monotonous interval Di' the end point position is the center, and the average value T of the gray scale values of R positions at the left and right sides at I is obtainedi；
Sixth, determine TiIf the number is largeAt 1, then with adjacent monotonic interval D1'、D2The average position of the end point position of' is the end position of the first threshold segmentation of the traversal, T1The threshold used for the first threshold segmentation is analogized in turn, and the starting position of the last threshold segmentation is an adjacent monotone interval Di-1'、Di' the average position of the end point position, the end position is the end position of the traversal, and the corresponding threshold is Ti(ii) a If the number is equal to 1, then only one threshold T is used for this traversal1；
Seventhly, searching 5 in each threshold segmentation range╳Pixels with a gray value greater than the corresponding threshold value in the 5 windows are assigned a value of 255, the 5 windows╳5, assigning 0 to the gray value of other pixel points in the column area covered by the window to realize the autonomous segmentation of the local threshold;
the second step is that: and performing nearest neighbor clustering on the two-dimensional data subjected to the local threshold autonomous segmentation, wherein the clustering distance threshold is n pixels, so that a certain number of data classes are obtained.
Further, the three prior probability density functions for distinguishing the target data from the interference data, which are designed in the second step, are specifically designed as follows:
the first step is as follows: the average values of the width in the vertical direction, the width uniformity and the compactness describing the weld position are w pixels, u pixels and cp pixels, respectively, and the variance is δ1One pixel, delta2A pixel sum delta3The prior probability density function of each pixel is described as formula (4):
wherein C is1Indicating the presence of a laser stripe of the type X1Representing a width characteristic attribute, X, of each class of data2Characteristic attribute representing width uniformity of each data class, X3Representing the width compactness characteristic attribute; and is
Is the average width of the ith class of data in the vertical direction,
is the average deviation of the width across the ith data class from the average width, where NiIs the number of coordinates in the horizontal direction, W, of the ith data classijIs the width of the ith class of data at the jth horizontal coordinate,
is the number of data in the jth column for the ith data class;
the second step is that: because the interference data is the opposite event of the welding seam position data, and according to the maximum posterior probability criterion, the prior probability density function of the interference data is designed to be small, and the probability density function is described as formula (5):
wherein C is2Indicates belonging to the category of interference data, and additionally promises P (C) under unknown interference conditions1)＝P(C2)。
Further, in the second step, the preliminary identification of the weld seam profile is realized based on a naive Bayes classification algorithm according to the maximum posterior probability criterion, and the specific steps are as follows:
the first step is as follows: calculating each class of data in the designation C1And C2Prior probability under the condition of (1);
the second step is that: judging the attribution type of each data class according to a maximum posterior probability rule, wherein the judgment rule that each data class belongs to the welding seam position is as follows:
P(X1|C1)P(X2|C1)P(X3|C1)＞P(X1|C2)P(X2|C2)P(X3|C2) (6)
if the formula (6) is satisfied, the data is welding seam position data, otherwise, the data is interference data.
Further, the third step comprises the following specific steps:
the first step is as follows: the spatial span of the kth data class is defined as:
wherein
Is the coordinates of the dissimilar data member of the kth data class in the image;
the second step is that: the fluctuation degree of the kth data class profile is defined as:
Mk＝N(sk·sk+1< 0) (8) wherein
Represents the slope, N (-) represents the arithmetic;
the third step: calculating visual characteristics Lk/MkThe larger it is, the higher the probability that the data class belongs to the weld position data is;
the fourth step: data class pairs L with overlapping coordinates in the horizontal directionk/MkSpreading out competition, Lk/MkThe large data class is regarded as weld position data, and interference is further removed.
Compared with the prior art, the invention has the beneficial effects that:
(1) the method can effectively overcome adverse effects caused by interference such as electric arc and splashing, can stabilize welding seam tracking in an automatic electric arc welding process based on laser vision sensing, and improves welding quality.
(2) An improved design method of a Gabor filter is provided, the filter can more effectively highlight the directional characteristic of a detected target than the traditional Gabor filter, and can provide reference for the detection of the directional characteristic of the target based on the Gabor filter.
(3) A naive Bayes classifier is designed, and the classifier can effectively identify the interference between the position of a welding line based on laser visual sensing and the background.
Drawings
FIG. 1 is a flow chart of weld position extraction based on a naive Bayes classifier;
FIG. 2 is an acquisition diagram of weld position comprehensive direction characteristics based on an improved Gabor filter; (a) is an original image; (b) a directional signature generated for the modified Gabor filter; (c) a direction feature map generated for a conventional Gabor filter;
FIG. 3 is an example of local threshold autonomous segmentation;
FIG. 4 is a classification diagram of data based on a naive Bayes classifier; (a) the result is the nearest neighbor clustering result; (b) the data classification result is obtained;
FIG. 5 is a diagram of a weld location extraction process; (a) the initial identification result of the welding seam position is obtained; (b) extracting a result for the weld position;
FIG. 6 is a processing result of extracting the weld positions of the butt joints according to the method of the present invention; (a) is an original image; (b) is a processing result graph;
FIG. 7 is a processing result of extracting the position of the weld of the lap joint according to the method of the present invention; (a) is an original image; (b) is a processing result graph.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in further detail below with reference to the accompanying drawings and embodiments. The embodiments described herein are only for explaining the technical solution of the present invention and are not limited to the present invention.
Example 1 autonomous extraction of T-Joint weld position
The extraction flow chart is shown in fig. 1, and comprises the following steps:
1. the test adopts a GMAW welding method, a T-shaped joint and a butt joint with the thickness of 50mm are selected, a robot is adopted for automatic welding to implement a welding process, the welding speed is 300.0mm/min, the wire feeding speed is 9000.0mm/min, and the welding current and voltage are automatically matched with the wire feeding speed;
2. collecting an original welding seam image, wherein the image comprises a complete electric arc area, a filling area, laser rays, a web plate and a bottom plate edge; the original image is shown in fig. 2 (a);
3. implementing direction characteristic detection based on the improved Gabor filter, respectively detecting a web plate, a groove and a welding seam region by using formulas (2) a-c, and detecting a bottom plate and a filtering angle theta by using a formula (2) b1～θ4Set to-10 °, -110 °, 5 °, and 30 °, respectively; the direction characteristic diagram generated by the improved Gabor filter of the present invention is shown in fig. 2(b), and the direction characteristic diagram generated by the conventional Gabor filter is shown in fig. 2(c), and it can be seen that the improved Gabor filter of the present invention can highlight the direction characteristic of the detected object more effectively than the conventional Gabor filter;
4. when the obtained direction feature map is subjected to local threshold autonomous segmentation, the lengths of two linear filters are set to be 23, the difference value G of gray values is 10 pixels, and R is 6; an example of local threshold autonomous segmentation is shown in fig. 3;
5. when nearest neighbor clustering is carried out on the two-dimensional data, a clustering distance threshold is set to be n-2 pixels; the nearest neighbor clustering result is shown in fig. 4 (a);
6. when data classification is implemented based on a naive Bayes classifier, in order to determine visual feature probability density functions of width, width uniformity and compactness of a welding seam position, 3 pixels are designed for w, 0.5 pixel for u, 0 pixel for cp and a variance delta1Is 0.4 pixel, delta2Is 0.4 pixel, delta30.5 pixels; the data classification result is shown in fig. 4 (b);
7. based on the maximum posterior probability criterion, performing the primary identification of the welding seam position; the result of the preliminary recognition of the weld position is shown in fig. 5 (a);
8. performing visual feature competition, and finally extracting weld contour data; the result of the bead position extraction is shown in fig. 5 (b).
Example 2 autonomous extraction of Butt Joint weld position
Applied welding position self-liftingThe procedure is similar to example 1, but the filter angle θ1～θ4Set at-1 °, -10 °, and 1 °, respectively (fig. 6).
Example 3 Lap Joint weld position autonomous extraction
The welding position autonomous extraction process implemented is similar to that of example 1, but with a filtering angle θ1～θ4Set at-2 °, -80 °, and 2 °, respectively (fig. 7).
The foregoing merely represents preferred embodiments of the invention, which are described in some detail and detail, and therefore should not be construed as limiting the scope of the invention. It should be noted that, for those skilled in the art, various changes, modifications and substitutions can be made without departing from the spirit of the present invention, and these are all within the scope of the present invention. Therefore, the protection scope of the present patent shall be subject to the appended claims.
Claims (6)
1. An autonomous extraction method of a welding seam position based on a naive Bayes classifier is characterized in that: the method comprises the following steps:
designing an improved Gabor filter to initially inhibit strong electric arc and splash interference of a welding seam image;
step two, performing local threshold autonomous segmentation on the filtered image, and realizing data classification according to a nearest neighbor clustering algorithm; three prior probability density functions of thickness, uniformity and compactness which are subject to normal distribution and are used for distinguishing target data and interference data are designed according to data classification, and preliminary identification of the welding seam outline is realized on the basis of a naive Bayes classification algorithm according to a maximum posterior probability criterion;
and step three, designing a visual feature calculation method of space span and fluctuation degree, performing overlapped unfolding visual feature competition on the preliminarily identified data in the horizontal direction, and finally accurately extracting the position of the welding seam.
2. The naive Bayes classifier-based autonomous extraction method of weld positions as claimed in claim 1, wherein: in the first step, an improved Gabor filter is designed to preliminarily inhibit strong electric arc and splash interference of a welding seam image, and the specific steps are as follows:
the first step is as follows: the improved construction method of the convolution kernel of the two-dimensional Gabor filter comprises the following steps:
where x '═ x cos θ + y sin θ, y' — x sin θ + y cos θ, f is the frequency in pixels, θ is the detection direction, σ is the convolution template size,
is a phase, designed specifically as
σ＝4，
n and m are designed as follows:
the second step is that: for the extraction of the welding seam position of the T-shaped joint, formulas (2) a-c are respectively used for detecting a web plate, a groove and a welding seam area, and formula (2) b is also used for detecting a bottom plate; set the filter angles to be theta1∈[-20°,-5°]、θ2∈[-80°,-110°]、θ3∈[5°,15°]And theta4∈[15°,35°](ii) a For the extraction of the position of the V-shaped groove weld of the butt joint, formulas (3) a-c are respectively used for detecting a plate surface and left and right groove areas;
the third step: and carrying out linear combination on the filtering results to generate a comprehensive direction characteristic diagram.
3. The naive bayes classifier-based autonomous extraction method of weld positions according to claim 2, characterized in that: in the second step, the filtered image is subjected to local threshold autonomous segmentation, and data classification is realized according to a nearest neighbor clustering algorithm, and the specific steps are as follows:
the first step is as follows: performing local threshold autonomous segmentation on the comprehensive direction characteristic graph, wherein the specific local threshold determination method comprises the following steps:
firstly, smoothing the image F in the vertical direction by adopting a 5 gamma 5 window with the step length of 5 to obtain average gray values at different height positions
Wherein I represents a row, j represents a column, and the vector I formed by average gray values obtained at each position is subjected to linear filtering twice, and the length of each filter is more than 19;
obtaining each monotone increasing interval of the filtered vector I', recording the starting position and the ending position of the monotone increasing interval, and simultaneously recording the gray value of the corresponding position;
calculating the difference value of the gray value at the two ends of each monotonous interval and recording the monotonous interval D with the difference value larger than G pixelsiThereby representing the gray value mutation area;
fourthly, calculating the monotonous interval DiThe difference value of the gray values of the vectors I and I' at each termination position is recorded again, and the monotone interval D with the difference value larger than G pixels is recorded againi', the abrupt change of the original gray value rising region;
using each monotonous interval Di' the end point position is the center, and the average value T of the gray scale values of R positions at the left and right sides at I is obtainedi；
Sixth, determine TiIf the number is greater than 1, with adjacent monotone intervals D1'、D2The average position of the end point position of' is the end position of the first threshold segmentation of the traversal, T1Threshold values for the first threshold segmentation, and so onThe initial position of the last threshold value division is an adjacent monotone interval Di-1'、Di' the average position of the end point position, the end position is the end position of the traversal, and the corresponding threshold is Ti(ii) a If the number is equal to 1, then only one threshold T is used for this traversal1；
Searching pixel points with gray values larger than corresponding thresholds in the 5 x 5 window in each threshold segmentation range, and assigning gray values of 255 to the pixel points and assigning gray values of 0 to other pixel points in the column area covered by the 5 x 5 window to realize local threshold independent segmentation;
the second step is that: and performing nearest neighbor clustering on the two-dimensional data subjected to the local threshold autonomous segmentation, wherein the clustering distance threshold is n pixels, so that a certain number of data classes are obtained.
4. The naive Bayes classifier-based autonomous extraction method of weld positions as claimed in claim 1, wherein: the three prior probability density functions for distinguishing the target data from the interference data, which are designed in the second step, are specifically designed as follows:
the first step is as follows: the average values of the width in the vertical direction, the width uniformity and the compactness describing the weld position are w pixels, u pixels and cp pixels, respectively, and the variance is δ1One pixel, delta2A pixel sum delta3The prior probability density function of each pixel is described as formula (4):
wherein C is1Indicating the presence of a laser stripe of the type X1Representing a width characteristic attribute, X, of each class of data2Characteristic attribute representing width uniformity of each data class, X3Representing the width compactness characteristic attribute; and is
Is the ith dataThe average width of the class in the vertical direction,
is the average deviation of the width across the ith data class from the average width, where NiIs the number of coordinates in the horizontal direction, W, of the ith data classijIs the width of the ith class of data at the jth horizontal coordinate,
is the number of data in the jth column for the ith data class;
the second step is that: because the interference data is the opposite event of the welding seam position data, and according to the maximum posterior probability criterion, the prior probability density function of the interference data is designed to be small, and the probability density function is described as formula (5):
wherein C is2Indicates belonging to the category of interference data, and additionally promises P (C) under unknown interference conditions1)＝P(C2)。
5. The naive Bayes classifier-based autonomous extraction method of weld positions as claimed in claim 4, wherein: in the second step, the primary recognition of the welding seam outline is realized based on a naive Bayes classification algorithm according to the maximum posterior probability criterion, and the specific steps are as follows:
the first step is as follows: calculating each class of data in the designation C1And C2Prior probability under the condition of (1);
the second step is that: judging the attribution type of each data class according to a maximum posterior probability rule, wherein the judgment rule that each data class belongs to the welding seam position is as follows:
P(X1|C1)P(X2|C1)P(X3|C1)＞P(X1|C2)P(X2|C2)P(X3|C2) (6)
if the formula (6) is satisfied, the data is welding seam position data, otherwise, the data is interference data.
6. The naive Bayes classifier-based autonomous extraction method of weld positions as claimed in claim 1, wherein: the third step comprises the following specific steps:
the first step is as follows: the spatial span of the kth data class is defined as:
wherein
Is the coordinates of the dissimilar data member of the kth data class in the image;
the second step is that: the fluctuation degree of the kth data class profile is defined as:
Mk＝N(sk·sk+1＜0) (8)
wherein
Represents the slope, N (-) represents the arithmetic;
the third step: calculating visual characteristics Lk/MkThe larger it is, the higher the probability that the data class belongs to the weld position data is;
the fourth step: data class pairs L with overlapping coordinates in the horizontal directionk/MkSpreading out competition, Lk/MkThe large data class is regarded as weld position data, and interference is further removed.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111066499.2A CN113762400B (en) | 2021-09-13 | 2021-09-13 | Autonomous extraction method for weld joint position based on naive Bayes classifier |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111066499.2A CN113762400B (en) | 2021-09-13 | 2021-09-13 | Autonomous extraction method for weld joint position based on naive Bayes classifier |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113762400A true CN113762400A (en) | 2021-12-07 |
| CN113762400B CN113762400B (en) | 2023-10-31 |
Family
ID=78795088
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111066499.2A Active CN113762400B (en) | 2021-09-13 | 2021-09-13 | Autonomous extraction method for weld joint position based on naive Bayes classifier |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113762400B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116823703A (en) * | 2023-02-03 | 2023-09-29 | 肇庆学院 | Structural laser weld image processing method based on Gabor filtering |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20110043033A (en) * | 2009-10-20 | 2011-04-27 | 삼성전기주식회사 | Metal pad state detection method using gabor filter |
| 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 |
| CN109003279A (en) * | 2018-07-06 | 2018-12-14 | 东北大学 | Fundus retina blood vessel segmentation method and system based on K-Means clustering labeling and naive Bayes model |
| US20190019589A1 (en) * | 2017-07-05 | 2019-01-17 | Electric Power Research Institute, Inc. | Apparatus and method for identifying cracks in a structure using a multi-stage classifier |
| CN110717872A (en) * | 2019-10-08 | 2020-01-21 | 江西洪都航空工业集团有限责任公司 | Method and system for extracting characteristic points of V-shaped welding seam image under laser-assisted positioning |
| CN110717531A (en) * | 2019-09-27 | 2020-01-21 | 华东师范大学 | Method for detecting classified change type based on uncertainty analysis and Bayesian fusion |
| CN111968072A (en) * | 2020-07-07 | 2020-11-20 | 南昌大学 | Thick plate T-shaped joint welding position autonomous decision-making method based on Bayesian network |
-
2021
- 2021-09-13 CN CN202111066499.2A patent/CN113762400B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20110043033A (en) * | 2009-10-20 | 2011-04-27 | 삼성전기주식회사 | Metal pad state detection method using gabor filter |
| 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 |
| US20190019589A1 (en) * | 2017-07-05 | 2019-01-17 | Electric Power Research Institute, Inc. | Apparatus and method for identifying cracks in a structure using a multi-stage classifier |
| CN109003279A (en) * | 2018-07-06 | 2018-12-14 | 东北大学 | Fundus retina blood vessel segmentation method and system based on K-Means clustering labeling and naive Bayes model |
| CN110717531A (en) * | 2019-09-27 | 2020-01-21 | 华东师范大学 | Method for detecting classified change type based on uncertainty analysis and Bayesian fusion |
| CN110717872A (en) * | 2019-10-08 | 2020-01-21 | 江西洪都航空工业集团有限责任公司 | Method and system for extracting characteristic points of V-shaped welding seam image under laser-assisted positioning |
| CN111968072A (en) * | 2020-07-07 | 2020-11-20 | 南昌大学 | Thick plate T-shaped joint welding position autonomous decision-making method based on Bayesian network |
Non-Patent Citations (2)
| Title |
|---|
| 何银水;杨杰;余卓骅;: "基于激光视觉系统的水下焊接焊缝识别", 电焊机, no. 11, pages 15 - 20 * |
| 余卓骅;胡艳梅;何银水;: "基于方向显著性的熔池背景下V形焊缝激光条纹的有效提取", 焊接, no. 01, pages 43 - 45 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116823703A (en) * | 2023-02-03 | 2023-09-29 | 肇庆学院 | Structural laser weld image processing method based on Gabor filtering |
| CN116823703B (en) * | 2023-02-03 | 2024-04-19 | 肇庆学院 | Structural laser weld image processing method based on Gabor filtering |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113762400B (en) | 2023-10-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108537808B (en) | A kind of gluing online test method based on robot teaching point information | |
| Muhammad et al. | Welding seam profiling techniques based on active vision sensing for intelligent robotic welding | |
| Li et al. | Robust welding seam tracking and recognition | |
| CN108765419B (en) | Structured light vision welding seam image information self-adaptive extraction method | |
| Wang | Monitoring of GMAW weld pool from the reflected laser lines for real-time control | |
| Muhammad et al. | A robust butt welding seam finding technique for intelligent robotic welding system using active laser vision | |
| CN106447669B (en) | The adhesion particle image concave point dividing method differentiated based on round masking-out area ratio | |
| CN111696107B (en) | Molten pool contour image extraction method for realizing closed connected domain | |
| CN103425988B (en) | Real-time positioning and matching method with arc geometric primitives | |
| CN108637435A (en) | A kind of three-dimensional seam tracking system and method for view-based access control model and arc voltage sensing | |
| CN110814465B (en) | Universal method for automatically extracting welding seam contour | |
| CN110717872B (en) | Method and system for extracting characteristic points of V-shaped welding seam image under laser-assisted positioning | |
| Li et al. | SVM-based information fusion for weld deviation extraction and weld groove state identification in rotating arc narrow gap MAG welding | |
| CN112365439B (en) | Method for synchronously detecting forming characteristics of GMAW welding seam of galvanized steel and direction of welding gun in real time | |
| Crouzil et al. | A new correlation criterion based on gradient fields similarity | |
| Lestriandoko et al. | Circle detection based on hough transform and Mexican Hat filter | |
| CN113674206B (en) | Extraction method suitable for characteristic parameters of deep-melting K-TIG welding molten pool and keyhole entrance | |
| CN106897990A (en) | The character defect inspection method of tire-mold | |
| CN113762400A (en) | Weld joint position autonomous extraction method based on naive Bayes classifier | |
| CN105689859A (en) | Connecting method and device for various metals | |
| CN108898621A (en) | A kind of Case-based Reasoning perception target suggests the correlation filtering tracking of window | |
| Kiddee et al. | Visual recognition of the initial and end points of lap joint for welding robots | |
| Lim et al. | Detection and estimation of circular arc segments | |
| CN113723494A (en) | Laser visual stripe classification and weld joint feature extraction method under uncertain interference source | |
| Lin et al. | Intelligent seam tracking of an ultranarrow gap during K-TIG welding: a hybrid CNN and adaptive ROI operation algorithm |
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 |