CN105931227B - A kind of face of weld defect characteristic extracting method for sweeping curve based on gradation of image B - Google Patents
A kind of face of weld defect characteristic extracting method for sweeping curve based on gradation of image B Download PDFInfo
- Publication number
- CN105931227B CN105931227B CN201610235624.0A CN201610235624A CN105931227B CN 105931227 B CN105931227 B CN 105931227B CN 201610235624 A CN201610235624 A CN 201610235624A CN 105931227 B CN105931227 B CN 105931227B
- Authority
- CN
- China
- Prior art keywords
- weld
- image
- edge
- line
- face
- 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.)
- Expired - Fee Related
Links
- 230000007547 defect Effects 0.000 title claims abstract description 51
- 238000000034 method Methods 0.000 title claims abstract description 47
- 238000010408 sweeping Methods 0.000 title claims abstract description 10
- 238000003466 welding Methods 0.000 claims abstract description 42
- 238000000605 extraction Methods 0.000 claims abstract description 22
- 239000011324 bead Substances 0.000 claims abstract description 6
- 238000001514 detection method Methods 0.000 claims description 38
- 238000001914 filtration Methods 0.000 claims description 17
- 238000012545 processing Methods 0.000 claims description 11
- 230000001788 irregular Effects 0.000 claims description 10
- 230000000694 effects Effects 0.000 claims description 9
- 239000000284 extract Substances 0.000 claims description 7
- 238000012360 testing method Methods 0.000 claims description 7
- 238000007781 pre-processing Methods 0.000 claims description 6
- 238000007689 inspection Methods 0.000 claims description 4
- 238000010606 normalization Methods 0.000 claims description 4
- 238000009825 accumulation Methods 0.000 claims description 2
- 238000006243 chemical reaction Methods 0.000 claims description 2
- 239000003086 colorant Substances 0.000 claims description 2
- 230000001186 cumulative effect Effects 0.000 claims description 2
- 230000009977 dual effect Effects 0.000 claims description 2
- 230000001629 suppression Effects 0.000 claims description 2
- 230000002463 transducing effect Effects 0.000 claims description 2
- 239000007787 solid Substances 0.000 claims 1
- 238000010586 diagram Methods 0.000 description 11
- 239000000047 product Substances 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 5
- 238000002474 experimental method Methods 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 208000003464 asthenopia Diseases 0.000 description 1
- 239000013066 combination product Substances 0.000 description 1
- 229940127555 combination product Drugs 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000003708 edge detection Methods 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 239000011521 glass Substances 0.000 description 1
- 238000009776 industrial production Methods 0.000 description 1
- 239000007788 liquid Substances 0.000 description 1
- 238000009659 non-destructive testing Methods 0.000 description 1
- 238000003908 quality control method Methods 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 238000011946 reduction process Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000005476 soldering Methods 0.000 description 1
- 239000004575 stone Substances 0.000 description 1
- 238000002604 ultrasonography Methods 0.000 description 1
- 238000011179 visual inspection Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
- G06T7/0008—Industrial image inspection checking presence/absence
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10024—Color image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20024—Filtering details
- G06T2207/20032—Median filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20048—Transform domain processing
- G06T2207/20061—Hough transform
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
- G06T2207/30152—Solder
Landscapes
- Engineering & Computer Science (AREA)
- Quality & Reliability (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
Miniature CCD camera acquisition parameter is arranged according to acquisition graphics standard in a kind of face of weld defect characteristic extracting method for sweeping curve based on gradation of image B;It converts collected true color image to gray-scale map, median filter process is carried out to image;It is tentatively extracted using edge line of the Canny operator edge line extraction algorithms between image inside weld region and background, between unwelded region and background based on threshold value;Judge the face of weld defects such as weldering partially, welding bead distortion according to edge line number and shape;Welded seam area continuous boundary line is reconstructed by Hough transform, realizes the accurate positionin of welded seam area;It draws and perpendicular to the section gray scale B of weld edge sweeps curve, when face of weld is there are when misalignment, B sweeps gray value difference of the curve at weld seam both ends and apparent variation occurs, the phenomenon that judging face of weld with this there are misalignments.The defects of realizing accurate positionin and inclined weldering, welding bead distortion, the misalignment of weld edge, accurately identifies.
Description
Technical field
The present invention relates to a kind of detection methods of face of weld defect, are based particularly on the weld defect of image processing techniques
Recognition methods.This method is identified suitable for automatic soldering technique face of weld defects detection, belongs to field of non destructive testing.
Background technology
Welding technique is extensively using in the industrial production, and the large-scale workpieces such as boiler, pipeline mostly use automatic welding work at present
Skill is completed.In producing wire type automatic Arc Welding, it inevitably will appear the misalignment for influencing product quality, weldering and disconnected weldering etc. partially lacks
It falls into.The presence of face of weld defect can substantially reduce safety of the welding product in service phase, product failure gently then occur and let out
Leakage, it is heavy then cause product generation brittle fracture, cause serious casualties.Therefore, for workpiece face of weld welding quality
Detection, for exclude workpiece major safety risks, ensure equipment safety operation provide reliable technical support.
Common welding detection means has visual detection, ray detection, ultrasound examination and Liquid penetrant testing etc., wherein visually
Detection is the most commonly used detection mode in the detection of weldment appearance.Visual detection is that welding products test personnel pass through magnifying glass, vernier
The detection instruments such as slide calliper rule and undercut measuring appliance, the test stone of combination product, the professional knowledge of welding products test personnel and detection
The quality condition of experience, butt-welding fitting surface is detected judge, to determine whether the component reaches design requirement.Visual inspection
Since its detection method flexibility is strong, simple operation and other advantages are widely applied.However, due to testing staff's long focus
Weld seam, eye strain, situations such as being susceptible to missing inspection, judge by accident.Further, since reviewer's quality, the difference of skills and experience,
There is deviation to the assurance of quality control standards (QCS) more, causes the Subjective Factors of the examined personnel of detection level larger, it is difficult to pair
The judgement that defect makes standardization, objectifies, standardizes and automate.
In recent years, for deficiency existing for above-mentioned conventional weld defect inspection method, developed and be based on image processing techniques
Weld defects detection recognition methods, for realizing the feature extraction of different type weld defect.Valavanis etc.
[Multiclass defect detection and classification in weld radio graphic images
using geometric and texture features[J].Expert Systems with Applications,
2010,37(12):7606-7614] for X-ray Images of Welding Seam carry out analysis and Study of recognition, it proposes to be based on digital picture
The algorithm of processing, handles the image collected, on the basis of ANNs Algorithm Analysis, obtains weld defect feature phase
Related parameter.[the An automatic system of classification of weld defects in such as Vilar
radiographic images[J].NDT&E International,2009,42(5):467-476] it is lacked for X-ray welding
Sunken image develops automatic defect classification system, which can realize image noise reduction and contrast enhancing, in conjunction with Da-Jin algorithm and mark
Label technology is split to welding defect image and the feature extraction of defect, is classified to defect by ANNs technologies, improves
To the efficiency and accuracy of defect recognition.
Currently, completing the technology of weld seam detection based on X-ray, mostly it is to be detected for weld seam welding internal flaw, closes
Also rarely has research in the image detection identification technology of face of weld defect.In this regard, method proposes one kind being based on Hough transform
The weld seam weld defects detection recognition methods of image processing techniques, for realizing the detection identification of the defect of face of weld.We
Method use Mechanical course miniature CCD camera automatic collection workpiece welded seam area image, using medium filtering, Canny operators and
Hough transform scheduling algorithm positions weld seam, meanwhile, by gray scale B sweep curve to the characteristic parameter of face of weld defect into
Row extraction.
Invention content
The purpose of the present invention is to provide a kind of detection recognition methods of weld seam weld defect, are based particularly on gradation of image
B sweeps the detection method of curve.Image processing techniques is used for weld seam detection by this method, using Hough transform edge line extraction skill
Art judges that weld seam with the presence or absence of being partially welded, sweeps curve by B and extracts weld seam to weld seam zone location, according to the number of extraction edge line
The characteristic parameter for the defects of there are misalignments on surface.
The present invention proposes a kind of face of weld defects detection recognition methods for sweeping curve based on gradation of image B, substantially former
Reason is:
Using the colored weld image of industrial CCD camera acquisition, the weld image of acquisition include face of weld exist be partially welded,
Misalignment defect.
To collected colored weld image gray processing processing, deposited between gray level image gray value and coloured image rgb value
Coloured image can be converted by gray level image according to the transformational relation in intrinsic transformational relation.Gray level image ash after conversion
Shown in transformational relation such as formula (1) between angle value Gray and original color image rgb value.
Gray=R*0.3+G*0.59+B*0.11 (1)
R, G and B indicates three kinds of colors of RGB respectively;Before handling weld seam, image need to be filtered, to subtract
Few influence of the ambient noise to image recognition, Comprehensive Correlation medium filtering, smothing filtering and sharp filtering are to weld beam shape
Effect, choose method of the medium filtering as image preprocessing.Medium filtering is a kind of Nonlinear harmonic oscillator method, can be real
The noise reduction process of existing image.This method divides an image into the sliding window containing odd number point, uses each pixel in window respectively
Intermediate value gray value is come to replace specified pixel gray value, specified pixel be window center point, specified pixel gray value mathematical formulae table
It is shown as:
yi=med { fi-v,···,fi-1,fi,fi+1,···,fi+v} (2)
Wherein i ∈ Z, v=(m-1)/2, yiFor the weld image after medium filtering, fiFor the image before medium filtering, m is
The odd point of sliding window.Medium filtering both can in turn avoid the fuzzy distortion of weld edge to weld seam image noise reduction, realize weldering
Stitch the equivalent recovery of image.
The region where weld seam is extracted using the method for edge extracting, because Canny operators can by the setting of threshold value,
Strong edge and weak edge are extracted, more can really reflect the actual conditions at weld seam both ends, and Canny operator edge extractings
Algorithm has preferable signal-to-noise ratio, high rim positioning performance and preferable detection result in a noisy environment, is suitable for different rings
Edge detection under border.The principle of Canny operators is as follows:(1) image is filtered with Gaussian filter, is removed in image
Noise;(2) image is filtered with the first differential of Gauss operator, obtains gradient intensity and the direction of each image;(3) right
Gradient carries out " non-maxima suppression ", i.e. the pixel of each region is compared with different vicinity points, determines part
Maximum value;(4) twice threshold is taken to gradient.It is higher with threshold value, ambient noise is removed, while also having lost useful weld edge
Based on the picture of information.Relatively low with threshold value, the image for retaining weld edge information comes the edge of attachment weld;It is final to obtain just
The weld edge line of step.
The pseudo-edge line that many background noises generate is contained in the weld edge line tentatively obtained, in order to reject pseudo-edge
The influence of line is obtained the ideal region of weld seam, the edge line of weld seam is reconstructed using Hough transform.Hough transform is root
The edge extracting method that strong point line dual thought proposes.In image space, the line intersected in the conllinear corresponding parameter space of point;
In parameter space, all straight lines for intersecting at same point have conllinear point to be corresponding to it in image space.Therefore,
Straight-line detection problem in image space is transformed into the point test problems in parameter space by Hough transform, by parameter sky
Between in carry out simple cumulative statistics and complete straight-line detection task in image space.If using orthogonal coordinates in parameter space
Linear equation operation in system can cause parameter accumulation calculating amount to increase when image space straight slope is infinitely great, calculate
Redundancy increases, and operation time is long.To solve this problem, using the linear equation in polar coordinate system, such as formula (3) institute
Show:
ρ=xcos θ+ysin θ (3)
According to formula (3), it is assumed that crossing every bit has the straight line in n direction, usual n=180 that it is straight to detect weld edge at this time
The angle precision of line is 1 °.(ρ, the θ) coordinate for calculating separately n straight line, obtains n coordinate points.If it is N number of to wait for that judgement point has,
Finally obtained (ρ, θ) coordinate has the angle that N*n, wherein θ are discrete, shares 180 values.If multi-point and common-line,
Must have collinear points θ values be θiWhen, ρ value approximately equals are in ρi, i.e., collinear points are in straight line (ρi, θi) on.In straight-line detection, such as
Fruit, which is more than the discrete point of certain amount, identical (ρ, θ) coordinate, then can determine where there is straight lines.In polar coordinate system
In figure, it will be apparent that joint then indicates the straight line that a Hough transform goes out.By Hough transform, weld seam both ends can be reconstructed
Straight line, realize the positioning of welded seam area.
In conjunction with Canny operator Boundary extracting algorithms, can extract out between welded seam area and background, between unwelded region and background
Edge line.For the good weld edge of welding quality there are two edge lines, the weld seam that welding bead deviates will appear a plurality of edge
Straight line welds its edge line of the weld seam of distortion as irregular curve.In addition, for face of weld there are misalignments the case where, by
In two edges, height differs, and shows as the difference of gray value in the picture.Therefore, extensive by Hough transform straight line restructing algorithm
The continuous edge line of multiple welded seam area takes negative gray value using the row pixel value perpendicular to weld edge as X-axis with normalization
Section of weld joint image B, which is, for Y-axis sweeps curve.In welded seam area (region in such as Fig. 8 and Fig. 9), the good weldering of welding quality
Seam both ends gray value difference exists with misalignment weld seam both ends gray value difference differs apparent.Accordingly, it can determine whether face of weld difference
The weld defect of type.
A kind of face of weld defect characteristic extracting method for sweeping curve based on gradation of image B proposed by the present invention be by with
What lower step was realized:
Step 1:Image Acquisition.
Using industrial CCD camera automatic collection face of weld it is intact and exist partially weldering, misalignment defect image, the image of acquisition
For True color RGB image, picture size, clarity, form and aspect and saturation degree immobilize.
Step 2:Image preprocessing.
It converts True color RGB image to gray level image according to formula (1).Median filter process is carried out to gray level image,
Realize that image is restored in the noise reduction of edge fidelity.
Step 3:Canny operator edge line extractions.
Using add the Canny operator edge line extraction algorithms of threshold value to extract between image inside weld region and background, it is unwelded
Edge line between region and background.
Step 4:Judge defect type according to edge line number and shape.
If welded seam area inward flange line number mesh is more than two, it is identified as welding situation partially, if welded seam area inward flange line zero
It dissipates, irregular curve condition occurs, be then identified as welding bead distorting event.If the number of welded seam area inward flange line is two,
Then tentatively it is judged as the good weld seam of welding quality.
Step 5:Hough transform reconstructs welded seam area continuous boundary line.
For being tentatively judged as the good weld seam of welding quality, due to the influence of background noise and background texture, cause to be permitted
The generation of more pseudo-edges, these pseudo-edges are more scattered and irregular, and weld edge is discontinuous, in order to eliminate pseudo-edge pair
The influence of detection is reconstructed weld edge using Hough transform, extracts continuous weld edge line, eliminates pseudo-edge line
Influence to detection.
Step 6:B sweeps curve and judges misalignment defect type.
According to gray level image, draws the transversal gray scale B perpendicular to weld seam path and sweep curve.Wherein, transversal pixel value is X
Axis, it is Y-axis that normalization, which takes negative gray value,.According to the weld edge that step 5 positions, determine that B sweeps the weld seam transversal area in curve
Domain.When face of weld is there are when misalignment, B sweeps gray value difference of the curve at weld seam both ends and apparent variation occurs, with
The phenomenon that this judges face of weld there are misalignments.
The present invention has the following advantages:1) Canny operator edge line extraction algorithms and Hough transform welded seam area is continuous
Edge line reconfiguration technique is extracted for weld edge, realizes the accurate positionin of weld edge.2) it proposes gray scale section B sweeping song
Line realizes effective extraction of welding seam misalignment weld defect feature for analyzing face of weld defect characteristic parameter.
Description of the drawings
There are misalignment schematic diagrames for Fig. 1 faces of weld.
Fig. 2 faces of weld occur being partially welded situation intention.
Schematic diagram before Fig. 3 (a) median filter process.
Schematic diagram after Fig. 3 (b) median filter process.
There is the edge line extraction effect diagram for being partially welded situation in Fig. 4 faces of weld.
There is the edge line extraction effect diagram of welding distorting event in Fig. 5 faces of weld.
The good edge line extraction effect diagram of Fig. 6 face of weld welding qualities.
Fig. 7 Hough transforms reconstruct schematic diagram to weld edge line.
The second best in quality weld seam both ends gray value difference (position one) effect diagram of Fig. 8 surface soldereds.
The second best in quality weld seam both ends gray value difference (position two) effect diagram of Fig. 9 surface soldereds.
There are weld seam both ends gray value difference (position one) effect diagrams of misalignment on the surfaces Figure 10.
There are weld seam both ends gray value difference (position two) effect diagrams of misalignment on the surfaces Figure 11.
Figure 12 is the implementing procedure figure of this method.
Specific implementation mode
With reference to specific experiment, the invention will be further described:
The sample image that this experiment selection face of weld is welded there are misalignment and partially is as experiment sample.
Step 1:Image Acquisition.
There is the defects of being partially welded, misalignment image using industrial CCD camera acquisition transducing devices internal weld seams surface, acquisition
Image is True color RGB image, picture size 480*360, form and aspect 4, saturation degree 100, contrast -4, and each value immobilizes.
Step 2:Image preprocessing.
It converts True color RGB image to gray level image according to formula (1).As shown in Figs. 1-2.In being carried out to gray level image
Value filtering processing realizes that image is restored in the noise reduction of edge fidelity.As shown in Fig. 3 (a) -3 (b).
Step 3:Canny operator edge line extractions.
Using add the Canny operator edge line extraction algorithms of threshold value to extract between image inside weld region and background, it is unwelded
Edge line between region and background.The threshold value of Canny operators is set as [0.1,0.3].
Step 4:Judge defect type according to edge line number and shape.
The number and shape for judging edge, when in welded seam area (center picture position) edge line number be more than two, then
It is identified as occurring welding situation partially, as shown in Figure 4.When (center picture position) edge line is scattered in welded seam area, appearance is irregular
Curve condition, then be identified as welding distortion the case where, as shown in Figure 5.When (center picture position) edge line in welded seam area
Number be two, then be tentatively judged as the good weld seam of welding quality, as shown in Figure 6.
Step 5:Hough transform reconstructs welded seam area continuous boundary line.
For being tentatively judged as the good weld seam of welding quality, due to the influence of background noise and background texture, cause to be permitted
The generation of more pseudo-edges, these pseudo-edges are more scattered and irregular, and weld edge is discontinuous, in order to eliminate pseudo-edge pair
The influence of detection is reconstructed weld edge using Hough transform, extracts continuous weld edge line, eliminates pseudo-edge line
Influence to detection.As shown in Figure 7.
Step 6:B sweeps curve and judges misalignment defect type.
According to gray level image, draws the transversal gray scale B perpendicular to weld seam path and sweep curve.Wherein, transversal pixel value is X
Axis, it is Y-axis that normalization, which takes negative gray value,.According to the weld edge that step 5 positions, determine that B sweeps the weld seam transversal area in curve
Domain.When face of weld is there are when misalignment, B sweeps gray value difference of the curve at weld seam both ends and apparent variation occurs, with
The phenomenon that this judges face of weld there are misalignments.From experiment, it can be seen that when the good welded seam area of welding quality, weld seam
The gray value difference at both ends is 10-2On the order of magnitude, as Figure 8-9.When there are misalignments for weld seam the case where, the gray scale at weld seam both ends
Value difference value is 10-1On the order of magnitude.As shown in figs. 10-11.
The implementing procedure figure of this method is as shown in figure 12.
It is the typical case of the present invention above, application of the invention is without being limited thereto.
Claims (3)
1. a kind of face of weld defect characteristic extracting method for sweeping curve based on gradation of image B, it is characterised in that:This method is logical
Cross following steps realization:
Step 1:Image Acquisition;
Using industrial CCD camera automatic collection face of weld it is intact and exist partially weldering, misalignment defect image, the image of acquisition is true
Color RGB image, picture size, clarity, form and aspect and saturation degree immobilize;
Step 2:Image preprocessing;
Convert True color RGB image to gray level image;Median filter process is carried out to gray level image, realizes that image is protected at edge
Noise reduction in the case of true restores;
Step 3:Canny operator edge line extractions;
Using add the Canny operator edge line extraction algorithms of threshold value to extract between image inside weld region and background, unwelded region
Edge line between background;
Step 4:Judge defect type according to edge line number and shape;
If welded seam area inward flange line number mesh is more than two, it is identified as welding situation partially, if welded seam area inward flange line is scattered, goes out
Existing irregular curve condition, then be identified as welding bead distorting event;If the number of welded seam area inward flange line is two, tentatively
It is judged as the good weld seam of welding quality;
Step 5:Hough transform reconstructs welded seam area continuous boundary line;
For being tentatively judged as the good weld seam of welding quality, due to the influence of background noise and background texture, many puppets are caused
The generation at edge, these pseudo-edges are more scattered and irregular, and weld edge is discontinuous, in order to eliminate pseudo-edge to detection
Influence, weld edge is reconstructed using Hough transform, extracts continuous weld edge line, eliminates pseudo-edge line to inspection
The influence of survey;
Step 6:B sweeps curve and judges misalignment defect type;
According to gray level image, draws the transversal gray scale B perpendicular to weld seam path and sweep curve;Wherein, transversal pixel value is X-axis, is returned
It is Y-axis that one change, which takes negative gray value,;According to the weld edge that step 5 positions, determine that B sweeps the weld seam transversal region in curve;When
There are when misalignment, B sweeps gray value difference of the curve at weld seam both ends and apparent variation occurs face of weld, is judged with this
Face of weld the phenomenon that there are misalignments.
2. a kind of face of weld defect characteristic extracting method for sweeping curve based on gradation of image B according to claim 1, this
Image processing techniques is used for weld seam detection by method, using Hough transform edge line extraction technology to weld seam zone location, according to
The number of extraction edge line judges weld seam with the presence or absence of being partially welded, and sweeping curve extraction face of weld by B, there are the spies of misalignment defect
Levy parameter;
It is characterized in that:
Using the colored weld image of industrial CCD camera acquisition, the weld image of acquisition include face of weld exist be partially welded, misalignment
Defect;
To collected colored weld image gray processing processing, exist between gray level image gray value and coloured image rgb value solid
There is transformational relation, according to the transformational relation, coloured image can be converted into gray level image;Gray level image gray value after conversion
Shown in transformational relation such as formula (1) between Gray and original color image rgb value;
Gray=R*0.3+G*0.59+B*0.11 (1)
R, G and B indicates three kinds of colors of RGB respectively;Before handling weld seam, image need to be filtered, Comprehensive Correlation
Medium filtering, smothing filtering and sharp filtering choose medium filtering as image preprocessing to the effect of weld beam shape
Method;Median filter process method divides an image into the sliding window containing odd number point, uses respectively in window in each pixel
Value gray value replaces specified pixel gray value, and specified pixel is window center point, specified pixel gray value mathematical formulae indicates
For:
yi=med { fi-v,···,fi-1,fi,fi+1,···,fi+v} (2)
Wherein i ∈ Z, v=(m-1)/2, yiFor the weld image after medium filtering, fiFor the image before medium filtering, m is sliding
The odd point of window;
The region where weld seam is extracted using the method for edge extracting, the principle of Canny operators is as follows:(1) gaussian filtering is used
Device is filtered image, removes the noise in image;(2) image is filtered with the first differential of Gauss operator, is obtained
The gradient intensity of each image and direction;(3) " non-maxima suppression " is carried out to gradient, i.e., the pixel of each region is different
Vicinity points be compared, determine local maximum;(4) twice threshold is taken to gradient;Higher with threshold value, removal background is made an uproar
Sound, while based on also having lost the picture of useful weld edge information;It is relatively low with threshold value, retain the figure of weld edge information
Edge as carrying out attachment weld;Finally obtain preliminary weld edge line;
The pseudo-edge line that many background noises generate is contained in the weld edge line tentatively obtained, using Hough transform butt welding
The edge line of seam is reconstructed;Hough transform is the edge extracting method proposed according to dotted line dual thought;In image space
In, conllinear point corresponds to the line intersected in parameter space;In parameter space, all straight lines of same point are intersected in image sky
Between in there is conllinear point to be corresponding to it;Therefore, the straight-line detection problem in image space is transformed into parameter sky by Hough transform
Between in point test problems, pass through and carry out the straight-line detection that simple cumulative statistics is completed in image space in parameter space and appoint
Business;If using the linear equation operation in orthogonal coordinate system in parameter space, when image space straight slope is infinitely great,
Parameter accumulation calculating amount can be caused to increase, computing redundancy degree increases, and operation time is long;To solve this problem, it is sat using pole
Linear equation in mark system, as shown in formula (3):
ρ=x cos θ+y sin θs (3)
According to formula (3), it is assumed that crossing every bit has the straight line in n direction, n=180 to detect the angle of weld edge straight line at this time
Precision is 1 °;(ρ, the θ) coordinate for calculating separately n straight line, obtains n coordinate points;If it is N number of to wait for that judgement point has, finally obtain
(ρ, θ) coordinate have the angle that N*n, wherein θ are discrete, share 180 values;If multi-point and common-line, must have conllinear
Point is θ in θ valuesiWhen, ρ value approximately equals are in ρi, i.e., collinear points are in straight line (ρi, θi) on;In straight-line detection, if it exceeds one
Fixed number purpose discrete point has identical (ρ, θ) coordinate, then can determine where there is straight lines;It is bright in polar coordinate system figure
Aobvious joint then indicates the straight line that a Hough transform goes out;By Hough transform, the straight line at weld seam both ends is reconstructed, is realized
The positioning of welded seam area;
In conjunction with Canny operator Boundary extracting algorithms, the side between welded seam area and background, between unwelded region and background can extract out
Edge line;For the good weld edge of welding quality there are two edge lines, the weld seam that welding bead deviates will appear a plurality of edge line,
Its edge line of the weld seam of distortion is welded as irregular curve;In addition, for face of weld there are misalignments the case where, due to both sides
Edge height differs, and shows as the difference of gray value in the picture;Therefore, weld seam is restored by Hough transform straight line restructing algorithm
The continuous edge line in region takes negative gray value as Y-axis using the row pixel value perpendicular to weld edge as X-axis using normalization
It is section of weld joint image B and sweeps curve;In welded seam area, the good weld seam both ends gray value difference of welding quality and misalignment weld seam
It is apparent to there is difference in both ends gray value difference;Accordingly, it can determine whether the different types of weld defect of face of weld.
3. a kind of face of weld defect characteristic extracting method for sweeping curve based on gradation of image B according to claim 2,
It is characterized in that:
Choose the sample image that face of weld is welded there are misalignment and partially;
Step 1:Image Acquisition;
It is partially welded using the surface presence of industrial CCD camera acquisition transducing devices internal weld seams, misalignment defect image, the image of acquisition is
True color RGB image, picture size 480*360, form and aspect 4, saturation degree 100, contrast -4, each value immobilize;
Step 2:Image preprocessing;
It converts True color RGB image to gray level image according to formula (1);Median filter process is carried out to gray level image, is realized
Image is restored in the noise reduction of edge fidelity;
Step 3:Canny operator edge line extractions;
Using add the Canny operator edge line extraction algorithms of threshold value to extract between image inside weld region and background, unwelded region
Edge line between background;The threshold value of Canny operators is set as [0.1,0.3];
Step 4:Judge defect type according to edge line number and shape;
The number and shape for judging edge, when welded seam area inward flange line number mesh be more than two, then be identified as occurring welding situation partially;
The case where when welded seam area inward flange line is scattered, there is irregular curve condition, being then identified as welding distortion;Work as welded seam area
The number of inward flange line is two, then is tentatively judged as the good weld seam of welding quality;
Step 5:Hough transform reconstructs welded seam area continuous boundary line;
For being tentatively judged as the good weld seam of welding quality, due to the influence of background noise and background texture, many puppets are caused
The generation at edge, these pseudo-edges are more scattered and irregular, and weld edge is discontinuous, in order to eliminate pseudo-edge to detection
Influence, weld edge is reconstructed using Hough transform, extracts continuous weld edge line, eliminates pseudo-edge line to inspection
The influence of survey;
Step 6:B sweeps curve and judges misalignment defect type;
According to gray level image, draws the transversal gray scale B perpendicular to weld seam path and sweep curve;Wherein, transversal pixel value is X-axis, is returned
It is Y-axis that one change, which takes negative gray value,;According to the weld edge that step 5 positions, determine that B sweeps the weld seam transversal region in curve;When
There are when misalignment, B sweeps gray value difference of the curve at weld seam both ends and apparent variation occurs face of weld, is judged with this
Face of weld the phenomenon that there are misalignments.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610235624.0A CN105931227B (en) | 2016-04-14 | 2016-04-14 | A kind of face of weld defect characteristic extracting method for sweeping curve based on gradation of image B |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610235624.0A CN105931227B (en) | 2016-04-14 | 2016-04-14 | A kind of face of weld defect characteristic extracting method for sweeping curve based on gradation of image B |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105931227A CN105931227A (en) | 2016-09-07 |
CN105931227B true CN105931227B (en) | 2018-11-02 |
Family
ID=56838197
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610235624.0A Expired - Fee Related CN105931227B (en) | 2016-04-14 | 2016-04-14 | A kind of face of weld defect characteristic extracting method for sweeping curve based on gradation of image B |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105931227B (en) |
Families Citing this family (29)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108181315B (en) * | 2017-11-25 | 2021-02-02 | 天津大学 | Image processing-based biscuit damage detection device and detection method |
CN108127238A (en) * | 2017-12-29 | 2018-06-08 | 南京理工大学 | The method that non-burnishing surface autonomous classification robot increases material forming |
CN108645865A (en) * | 2018-05-10 | 2018-10-12 | 中国石油天然气集团有限公司 | A kind of measurement method of the submerged-arc welding steel pipe weld seam amount of the being partially welded parameter based on CCD |
CN108876771B (en) * | 2018-06-04 | 2022-05-20 | 广东工业大学 | Undercut welding defect detection method |
CN109540903A (en) * | 2018-11-28 | 2019-03-29 | 合肥常青机械股份有限公司 | Automated production procedure quality intelligent diagnosing method based on pattern-recognition |
CN111354039B (en) * | 2018-12-20 | 2023-07-14 | 核动力运行研究所 | Quick algorithm for extracting weld joint region based on B-scan image recognition |
CN109816652B (en) * | 2019-01-25 | 2022-08-12 | 湖州云通科技有限公司 | Complex casting defect identification method based on gray level significance |
CN109978865A (en) * | 2019-03-28 | 2019-07-05 | 中核建中核燃料元件有限公司 | A kind of method, apparatus for the detection of nuclear fuel rod face of weld |
CN110285908B (en) * | 2019-05-28 | 2022-04-01 | 北京机科国创轻量化科学研究院有限公司 | Real-time online detection method for multi-beam fiber tension in composite material preform manufacturing process |
CN110852173B (en) * | 2019-10-15 | 2022-08-09 | 山东大学 | Visual positioning method and system for fuzzy weld joint |
CN111047561A (en) * | 2019-11-22 | 2020-04-21 | 国网江西省电力有限公司电力科学研究院 | Method for identifying cracks of shed tortoise of composite insulator |
CN110910373B (en) * | 2019-11-25 | 2023-01-24 | 西南交通大学 | Identification method of orthotropic steel bridge deck fatigue crack detection image |
CN112001906B (en) * | 2020-08-24 | 2024-02-02 | 徐工汉云技术股份有限公司 | Steel plate weld image detection method combined with non-maximum suppression |
CN112258521B (en) * | 2020-10-16 | 2022-06-07 | 江苏方天电力技术有限公司 | Three-dimensional reconstruction method for pipeline weld joint based on single elliptical imaging radiograph |
CN112529884B (en) * | 2020-12-17 | 2022-03-25 | 中国石油大学(华东) | Welding spot quality evaluation method based on indentation characteristic image recognition |
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 |
CN113112470B (en) * | 2021-04-08 | 2023-12-15 | 揭阳市柏亿不锈钢有限公司 | Cold rolling process steel bending detection analysis method based on artificial intelligence |
CN113570548B (en) * | 2021-06-29 | 2022-06-03 | 北京工业大学 | Linear array camera-based gapless weld joint identification method and system |
CN113375850A (en) * | 2021-07-26 | 2021-09-10 | 浙江省冶金研究院有限公司 | Residual stress distribution testing method based on micro-indentation method |
CN113781482B (en) * | 2021-11-11 | 2022-02-15 | 山东精良海纬机械有限公司 | Method and system for detecting crack defects of mechanical parts in complex environment |
CN115338556A (en) * | 2022-08-03 | 2022-11-15 | 湖南科技大学 | Weld joint quality detection method for thick-wall welding workpiece and computer equipment |
CN115511819B (en) * | 2022-09-21 | 2023-08-25 | 武汉理工大学 | X-ray weld boundary identification method, device, equipment and storage medium |
CN115272307B (en) * | 2022-09-26 | 2022-12-30 | 无锡日联科技股份有限公司 | Defect detection method and system for alloy type temperature fuse |
CN116503408B (en) * | 2023-06-28 | 2023-08-25 | 曲阜远大集团工程有限公司 | Scanning technology-based steel structure surface defect detection method |
CN116630312B (en) * | 2023-07-21 | 2023-09-26 | 山东鑫科来信息技术有限公司 | Visual detection method for polishing quality of constant-force floating polishing head |
CN116935039B (en) * | 2023-09-15 | 2023-12-29 | 深圳市泽信智能装备有限公司 | New energy battery welding defect detection method based on machine vision |
CN117541582B (en) * | 2024-01-09 | 2024-04-19 | 山东海纳智能装备科技股份有限公司 | IGBT insulation quality detection method for high-frequency converter |
CN117541588B (en) * | 2024-01-10 | 2024-03-12 | 大连建峰印业有限公司 | Printing defect detection method for paper product |
CN117576106B (en) * | 2024-01-17 | 2024-03-29 | 高科建材(咸阳)管道科技有限公司 | Pipeline defect detection method and system |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104036495A (en) * | 2014-05-22 | 2014-09-10 | 南京航空航天大学 | Welding defect extraction method and welding defect detection method |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP6031339B2 (en) * | 2012-11-21 | 2016-11-24 | 富士フイルム株式会社 | Perspective image density correction method, nondestructive inspection method, and image processing apparatus |
-
2016
- 2016-04-14 CN CN201610235624.0A patent/CN105931227B/en not_active Expired - Fee Related
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104036495A (en) * | 2014-05-22 | 2014-09-10 | 南京航空航天大学 | Welding defect extraction method and welding defect detection method |
Non-Patent Citations (2)
Title |
---|
Multiclass defect detection and classification in weld radiographic images using geometric and texture features;Ioannis Valavanis et al;《Expert Systems with Applications》;20101231;第37卷(第12期);7606-7614 * |
基于Canny算子的焊缝图像边缘提取技术;谢志孟 等;《焊接学报》;20060131;第27卷(第1期);29-36 * |
Also Published As
Publication number | Publication date |
---|---|
CN105931227A (en) | 2016-09-07 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105931227B (en) | A kind of face of weld defect characteristic extracting method for sweeping curve based on gradation of image B | |
CN105976352B (en) | One kind being based on the morphologic face of weld defect characteristic extracting method of gray level image | |
CN108665452B (en) | Pipeline-weld film scanning storage and identification of Weld Defects and its system | |
Shafeek et al. | Automatic inspection of gas pipeline welding defects using an expert vision system | |
CN105913415B (en) | A kind of image sub-pixel edge extracting method with extensive adaptability | |
CN102175700B (en) | Method for detecting welding seam segmentation and defects of digital X-ray images | |
US20140348415A1 (en) | System and method for identifying defects in welds by processing x-ray images | |
WO2021251064A1 (en) | Information processing device, determination method, and information processing program | |
CN109886298A (en) | A kind of detection method for quality of welding line based on convolutional neural networks | |
Tao et al. | Weak scratch detection and defect classification methods for a large-aperture optical element | |
CN111696107A (en) | Molten pool contour image extraction method for realizing closed connected domain | |
Islam et al. | Capsule image segmentation in pharmaceutical applications using edge-based techniques | |
CN110717909A (en) | Metal surface scratch detection method and device | |
CN107154033A (en) | A kind of high ferro contact net rotation ears vertical openings pin missing detection method and system | |
CN115836218A (en) | Inspection device, inspection method, and inspection program | |
Jeon et al. | Non-destructive photoacoustic imaging of metal surface defects | |
CN103177426B (en) | Morphology-based strong-interference laser edge image restoration method | |
Dang et al. | A novel method for detecting weld defects accurately and reliably in radiographic images | |
Hashmi et al. | Computer-vision based visual inspection and crack detection of railroad tracks | |
Laghari et al. | Identification of faulty BGA solder joints in X-ray images | |
Thiruganam et al. | Automatic defect detection and counting in radiographic weldment images | |
Louban | Image processing of edge and surface defects | |
CN114820493B (en) | Method for detecting split of composite material orifice caused by hole making | |
Islam et al. | Computer vision-based quality inspection system of transparent gelatin capsules in pharmaceutical applications | |
Zheng et al. | Measurement of laser welding pool geometry using a closed convex active contour model |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20181102 |
|
CF01 | Termination of patent right due to non-payment of annual fee |