CN104730091B - The extraction of gas turbine blades defect and analysis method based on region segmentation detection - Google Patents
The extraction of gas turbine blades defect and analysis method based on region segmentation detection Download PDFInfo
- Publication number
- CN104730091B CN104730091B CN201510068469.3A CN201510068469A CN104730091B CN 104730091 B CN104730091 B CN 104730091B CN 201510068469 A CN201510068469 A CN 201510068469A CN 104730091 B CN104730091 B CN 104730091B
- Authority
- CN
- China
- Prior art keywords
- mtd
- mrow
- image
- mtr
- msub
- 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
Landscapes
- Analysing Materials By The Use Of Radiation (AREA)
Abstract
Gas turbine blades are first divided into some regions by the extraction of gas turbine blades defect and analysis method based on region segmentation detection using galvanized wire, are carried out subregion flaw detection using DR equipment, are obtained blade overall intensity image A；Then galvanized wire displacement detect a flaw inside subregion, obtain blade overall intensity image B；Image A, B progress image co-registration are obtained into the overall X ray gray scale original image for being free of galvanized wire；Then X ray gray scale original image is pre-processed, obtains the X-ray detection image of blade；X-ray detection image is handled again to obtain the binary image of defect protrusion, two-dimensional coordinate system is established on binary image, positioning analysis is carried out to existing internal flaw；Finally calculate the girth sum of all pixels and area sum of all pixels of defect in binary image, the defects of actual size is obtained by conversion, qualitative and quantitative analysis is carried out to existing internal flaw, the present invention realizes the fast and automatically extraction and analysis to blade interior defect.
Description
Technical field
The invention belongs to technical field of nondestructive testing, and in particular to the gas turbine blades defect based on region segmentation detection
Extraction and analysis method.
Background technology
Gas turbine is a kind of is converted to machinery using the fuel-air mixture body that continuously flows as working medium, heat energy
The rotary power machinery of work(.Gas turbine is high efficiency of energy conversion and clean utilization system in 21 century or even longer period
Core drive is equipped.At present, China is in the manufacture that the large-sized powers such as gas turbine are equipped, the capability of independent innovation is weak, externally according to
Degree of depositing height.Therefore, the research work in this field is carried out, to the technology for breaking through correlation technique bottleneck problem, breaking developed country
Block has important and far-reaching strategic importance, closely bound up with the development of national economy.
Important component of the gas turbine turbine high temperature blade as gas turbine, from nickel-base high-temperature alloy material,
The manufacture of complicated airfoil is realized using the method for hot investment casting, the internal flaws such as crackle, shrinkage cavity can be produced, can direct shadow
The military service intensity and service life of blade are rung, and then directly affects the running stability of whole gas turbine and its using the longevity
Life.Traditional gas turbine turbine blade detection be using the method for radiographic film photograph, have the advantages that it is intuitive and reliable, thus
It is widely used in playing an important role in industrial production and in terms of quality control.But this method exist workload it is big,
Operating cost is high, detection process is complicated and evaluation result has the deficiencies of certain subjectivity.
The content of the invention
The shortcomings that in order to overcome above-mentioned prior art, it is an object of the invention to provide the combustion gas detected based on region segmentation
Turbine blade defect is extracted and analysis method, it is possible to achieve fast and automatically extraction and analysis to blade interior defect.
The extraction of gas turbine blades defect and analysis method based on region segmentation detection, comprise the following steps：
1) according to the size of blade, thickness Curvature varying, blade is divided into more than two regions using galvanized wire, profit
Gas turbine blades are carried out with digital radial equipment (DR) to detect a flaw inside subregion, to the image that flaw detection obtains using white galvanized wire as
Feature carries out image mosaic, obtains the blade overall intensity image A for including galvanized wire；
2) galvanized wire is moved, avoids the position where galvanized wire during subregion first, do two subzones, same is divided into blade
With the equal number of region of step 1), gas turbine blades are carried out using digital radial to detect a flaw inside subregion, flaw detection is obtained
Image characterized by white galvanized wire carry out image mosaic, obtain the blade overall intensity image B for including galvanized wire；
3) by image A and image B on the basis of blade entirety common first edges feature, carry out image co-registration and obtain overall be free of
The X ray gray scale original image of galvanized wire；
4) X ray gray scale original image is pre-processed, obtains the X-ray detection image of blade；
5) X-ray detection image is handled to obtain the binary image of defect protrusion, established on binary image
Two-dimensional coordinate system, positioning analysis is carried out to existing internal flaw；
6) the girth sum of all pixels and area sum of all pixels of defect in binary image are calculated, reality is obtained by conversion
The defects of size, quantitative analysis is carried out to existing internal flaw, passes through and related feature --- axial ratio is extracted to defect
(the ratio between the major axis of defect area, short axle), girth area ratio, are classified based on the feature extracted to defect, so as to realize
Qualitative analysis to blade interior defect.
What described step 1) image A was obtained concretely comprises the following steps：
1.1) exposure parameter (including electric current, voltage, the exposure using X-ray machine to the Ith subregion of blade according to the Ith subregion
Time) 6 flaw detections are carried out, obtain the X-ray detection original image of 6 the Ith subregions；
1.2) by controlling rotatable stage that X-ray machine and DDA are visited to the blade anglec of rotation, control XYZ guide rail motions
Survey device movement so that the central point of x-ray source faces the central point of the subregion of blade the IIIth, to the IIIth subregion of blade according to the
The exposure parameter of III subregion carries out 6 flaw detections, obtains the X-ray detection original image of 6 the IIIth subregions；
1.3) same operation, obtain the V, the VI, IV, the X-ray detection original image of II subregion；
1.4) the 6 width original images obtained on each subregion, which make 6 average Removing Random Nos, to be influenceed；
1.5) characteristic point relevant to the image zooming-out white galvanized wire obtained after 1.4) handling, is calculated according to formula formula (1)
Image translation rotation transformation coordinate relative between any two is obtained, realizes image registration, i.e.,：
Assuming that common ground is respectively (x between adjacent two images11,y11)、(x12,y12)、(x21,y21)、(x22,y22) by
Following transformation for mula (1) can try to achieve transformation parameter cos θ, sin θ, tx、ty,
1.6) splicing is carried out to 6 width images based on the result after registration, schemed after determining splicing according to formula (2)-formula (6)
The size of picture：Wherein w1、h1、w2、h2For the width of two images to be spliced, height；
xmin=min (Aa+b(1,:)0) (3)
xmax=max (Aa+b(1,:)w1) (4)
ymin=min (Aa+b(2,:)0) (5)
ymax=max (Aa+b(2,:)h1) (6)
1.7) by after image rotation translation transformation registration to be spliced, spliced map is treated using formula (7) bilinear interpolation
As carrying out interpolation resampling, preliminary spliced image is obtained；
V (x, y)=ax+by+cxy+d (7)
The unknown equation that wherein 4 coefficients a, b, c, d are write out by 4 (x, y) adjoint points determines,
1.8) later stage smoothing processing is carried out to spliced image, changed using 3 × 3 template smooth grey of formula (8)
Prominent splicing line, obtain the blade overall intensity image A for including galvanized wire；
Described step 5) defect location analysis comprises the following steps that：
5.1) the X-ray detection image of integral blade is smoothed with the average value template of 3 × 3 pixels,
Simulate flawless image；
5.2) flawless analog image and overall X-ray detection image are made into difference operation, obtains defect protrusion
Error image；Because the existing gray value of defect area is the black region near 0, it is the white area near 255 to have gray value again
Domain, so needing to do the image that difference operation twice obtains black defect area protrusion and White Defects region protrudes respectively；
5.3) error image protruded defect makees Otsu automatic threshold segmentation computings and obtains containing defective binary picture
Picture；
5.4) binary image that black defect area protrudes and White Defects region protrudes is synthesized into computing, and does region
Growth, obtains the binaryzation defect image of the complete fidelity in region；
5.5) rectangular coordinate system as the origin of coordinates, is established in the upper left corner in gas turbine blades region using on image；
5.6) transverse and longitudinal coordinate of each defect is tried to achieve with the coordinate system of foundation；
5.7) the transverse and longitudinal coordinate X by 5.6) obtainingn,Yn；To Xn,YnMake computing, the center of gravity for calculating each defect is sat
Mark, is realized as follows to the positioning analysis center of gravity calculation formula of defect；
Described step 6) Flaw discrimination analysis comprises the following steps that：
6.1) the X detection images of integral blade are subjected to rim detection using single order edge detection operator；
6.2) opening operation is made to the image after rim detection, eliminates the influence of noise spot；
6.3) by the image after step 6.2) processing with doing add operation containing defective bianry image in step 5.3),
Eliminate the template smoothly influence to defect image；
6.4) image obtained to step 6.3) detects with a second order edge detective operators to it；
6.5) image in step 6.4) is calculated into the wherein Euclidean distance between pixel two-by-two, finds out maximum therein
The as most major axis of the defect image, and calculate the pixel transverse and longitudinal coordinate at most major axis both ends；
6.6) slope K is tried to achieve by the transverse and longitudinal coordinate obtained in step 6.5)；
6.7) calculation procedure 6.4) in image straight slope is the pixel set of (- 1/K) between pixel two-by-two；
6.8) pixel set for obtaining step 6.7), Euclidean distance corresponding to calculating, maximum therein is found out, is
The minor axis length of the defect；
6.9) the girth area ratio in axial ratio and step 5.7) is calculated；
6.10) by obtaining major and minor axis length ratio and girth area ratio, the common type for determining defect, i.e. major and minor axis
When length ratio, girth area ratio are all higher than the threshold value each given, defect type is crack defect, is otherwise shrinkage cavity class
Defect, so it is achieved that the qualitative analysis to defect.
Beneficial effects of the present invention：Subregion flaw detection is carried out to gas turbine blades using digital radial equipment, by lacking
Extraction correlated characteristic is fallen into, defect is classified based on the feature extracted, so as to realize positioning, calmly to blade interior defect
Property and quantitative analysis, overcome that radiographic film photographic process workload is big, operating cost is high, detection process is complicated and judge
As a result the deficiencies of certain subjectivity be present.
Brief description of the drawings
Fig. 1 is the subzone schematic diagram of gas turbine blades one.
Fig. 2 is the subzone schematic diagram of gas turbine blades two.
Fig. 3 is image homomorphic filtering flow chart.
Fig. 4 is segmentation contrast stretching schematic diagram.
Fig. 5 is blade defect extraction system schematic diagram.
Fig. 6 is that blade defect obtains analysis software schematic diagram.
Embodiment
The present invention is described in detail with reference to the accompanying drawings and examples.
The extraction of gas turbine blades defect and analysis method based on region segmentation detection, comprise the following steps：
1) due to gas turbine blades shape scrambling, the inhomogeneities of thickness and the size of flat panel detector
Limitation, it is necessary to the position different to gas turbine blades is detected a flaw using different transmission powers, different time for exposure,
Need to carry out subregion flaw detection to blade, each subregion needs to be detected a flaw using suitable exposure parameter, exposure less than normal bigger than normal
Optical parameter can all influence the image quality of digital radial, so being first according to the size of blade, thickness Curvature varying utilizes
Blade is divided into 6 regions and carries out subregion flaw detection by galvanized wire, as shown in figure 1, black dotted lines are the position where galvanized wire, due to galvanized wire
X ray can be absorbed well, white mutation is shown as in gray level image, so the feature as image mosaic, utilizes number
Word ray equipment is carried out detecting a flaw inside subregion to gas turbine blades, and the image obtained to flaw detection is carried out characterized by white galvanized wire
Image mosaic, obtain the blade overall intensity image A for including galvanized wire；
Comprise the following steps that：
1.1) exposure parameter (including electric current, voltage, the exposure using X-ray machine to the Ith subregion of blade according to the Ith subregion
Time) 6 flaw detections are carried out, obtain the X-ray detection original image of 6 the Ith subregions；
1.2) by controlling rotatable stage that X-ray machine and DDA are visited to the blade anglec of rotation, control XYZ guide rail motions
Survey device movement so that the central point of x-ray source faces the central point of the subregion of blade the IIIth, to the IIIth subregion of blade according to the
The exposure parameter of III subregion carries out 6 flaw detections, obtains the X-ray detection original image of 6 the IIIth subregions；
1.3) same operation, obtain the V, the VI, IV, the X-ray detection original image of II subregion；
1.4) the 6 width original images obtained on each subregion, which make 6 average Removing Random Nos, to be influenceed；
1.5) characteristic point relevant to the image zooming-out white galvanized wire obtained after 1.4) handling, is calculated according to formula formula (1)
Image translation rotation transformation coordinate relative between any two is obtained, realizes image registration, i.e.,：
Assuming that common ground is respectively (x between adjacent two images11,y11)、(x12,y12)、(x21,y21)、(x22,y22) by
Following transformation for mula (1) can try to achieve transformation parameter cos θ, sin θ, tx、ty,
1.6) splicing is carried out to 6 width images based on the result after registration, schemed after determining splicing according to formula (2)-formula (6)
The size of picture：Wherein w1、h1、w2、h2For the width of two images to be spliced, height；
xmin=min (Aa+b(1,:)0) (3)
xmax=max (Aa+b(1,:)w1) (4)
ymin=min (Aa+b(2,:)0) (5)
ymax=max (Aa+b(2,:)h1) (6)
1.7) by after image rotation translation transformation registration to be spliced, spliced map is treated using formula (7) bilinear interpolation
As carrying out interpolation resampling, preliminary spliced image is obtained；
V (x, y)=ax+by+cxy+d (7)
The unknown equation that wherein 4 coefficients a, b, c, d are write out by 4 (x, y) adjoint points determines,
1.8) later stage smoothing processing is carried out to spliced image, changed using 3 × 3 template smooth grey of formula (8)
Prominent splicing line, obtain the blade overall intensity image A for including galvanized wire；
2) in order to prevent galvanized wire from causing to block to defect, galvanized wire is moved, avoids the position where galvanized wire during subregion first,
Two subzones are done, as shown in Fig. 2 same is divided into blade in 6 regions, gas turbine blades are divided using digital radial
Detected a flaw inside area, the image obtained to flaw detection carry out image mosaic, repeat step 1.1 characterized by white galvanized wire)-step
1.8) the blade overall intensity image B for including galvanized wire, is obtained；
3) by image A and image B on the basis of blade entirety common first edges feature, carry out image co-registration and obtain overall be free of
The X ray gray scale original image of galvanized wire；
Specially：According to formula (9) to 1.8) image after processing carries out Hough transformation, and to obtain galvanized wire in image A straight for detection
Position coordinates where line；
ρ=xcos θ+ysin θ (9)
Image A and image B are subjected to coordinate system unification on the basis of blade entirety common first edges feature, will be examined in image A
The gray value at galvanized wire coordinate is measured to be substituted to realize the fusion of image by the gray value of corresponding position in image B, and then
Obtain the overall X ray blade gray level image for being free of galvanized wire；
4) X ray gray scale original image is pre-processed, obtains the X-ray detection image of blade；
The pretreatment of X ray gray scale original image comprises the following steps：
4.1) because DR equipment radiographic source is not strict spot light, ray image is inevitably by the shadow scattered
Ring, this partial dispersion noise causes the fuzzy of ray image, reduces the contrast of image, after average to each subregion 6 times
Image make homomorphic filtering, eliminate the influence of shot noise, image homomorphic filtering flow is as shown in Figure 3；
4.2) due to the radioscopic image dark collected, so the increased processing of contrast need to be done to image：Due to
It is 0~255 all gray areas that blade defect area, which contains gray value, and background parts only include some middle section
Grayscale portion, power law conversion contrast stretching is carried out to background parts according to formula (10) and then realizes the effect of prominent defect part
Fruit, segmentation contrast stretching as shown in figure 4,
S=crg (10)
5) X-ray detection image is handled to obtain the binary image of defect protrusion, established on binary image
Two-dimensional coordinate system, positioning analysis is carried out to existing internal flaw；
Comprise the following steps that：
5.1) the X-ray detection image of integral blade is smoothed with the average value template of 3 × 3 pixels,
Simulate flawless image；
5.2) flawless analog image and overall X-ray detection image are made into difference operation, obtains defect protrusion
Error image；Because the existing gray value of defect area is the black region near 0, it is the white area near 255 to have gray value again
Domain, so needing to do the image that difference operation twice obtains black defect area protrusion and White Defects region protrudes respectively；
5.3) error image protruded defect makees Otsu automatic threshold segmentation computings and obtains containing defective binary picture
Picture；
5.4) binary image that black defect area protrudes and White Defects region protrudes is synthesized into computing, and does region
Growth, obtains the binaryzation defect image of the complete fidelity in region；
5.5) rectangular coordinate system as the origin of coordinates, is established in the upper left corner in gas turbine blades region using on image；
5.6) transverse and longitudinal coordinate of each defect is tried to achieve with the coordinate system of foundation；
5.7) the transverse and longitudinal coordinate X by 5.6) obtainingn,Yn；To Xn,YnMake computing, the center of gravity for calculating each defect is sat
Mark, is realized as follows to the positioning analysis center of gravity calculation formula of defect；
6) the girth sum of all pixels and area sum of all pixels of defect in binary image are calculated, reality is obtained by conversion
The defects of size, to existing internal flaw carry out quantitative analysis；
The boundary pixel number n for trying to achieve each defect is calculated by boundary tracking method, is tried to achieve by area pixel calculating method
The pixel count m that defect area includes, carry out conversion and try to achieve the quantitative analysis of the girth of defect area, area realization to defect,
When carrying out qualitative analysis to defect, automatic detection need to be carried out to the major and minor axis of defect, comprised the following steps that：
6.1) the X detection images of integral blade are subjected to rim detection using single order edge detection operator；
6.2) opening operation is made to the image after rim detection, eliminates the influence of noise spot；
6.3) by the image after step 6.2) processing with doing add operation containing defective bianry image in step 5.3),
Eliminate the template smoothly influence to defect image；
6.4) image obtained to step 6.3) detects with a second order edge detective operators to it；
6.5) image in step 6.4) is calculated into the wherein Euclidean distance between pixel two-by-two, finds out maximum therein
The as most major axis of the defect image, and calculate the pixel transverse and longitudinal coordinate at most major axis both ends；
6.7) slope K is tried to achieve by the transverse and longitudinal coordinate obtained in step 6.5)；
6.7) calculation procedure 6.4) in image straight slope is the pixel set of (- 1/K) between pixel two-by-two；
6.8) pixel set for obtaining step 6.7), Euclidean distance corresponding to calculating, maximum therein is found out, is
The minor axis length of the defect；
6.10) the girth area ratio in axial ratio and step 5.7) is calculated；
6.10) by obtaining major and minor axis length ratio and girth area ratio, the common type for determining defect, i.e. major and minor axis
When length ratio, girth area ratio are all higher than the threshold value each given, defect type is crack defect, is otherwise shrinkage cavity class
Defect, so it is achieved that the qualitative analysis to defect.
Used in the present invention defect extraction with analysis system as shown in figure 5, the system hardware mainly have digital radial equipment 4,
Computer 1, realize the data transmission unit 3 of data communication therebetween and the most of composition of controller 2 four, the system core
Part is computer 1.The composition frame chart of the software of the system is controlled as shown in fig. 6, when image obtains by the machinery of computer 1
Partial software processed passes through opening and closing, displacement, the rotation of turntable of the controller 2 to the X-ray machine and DDA detectors of digital ray equipment 4
Gyration is controlled；The image obtained to detection, the inside of computer 1 is transferred to by signal transmission unit 3, then by image
Manage software and Treatment Analysis is carried out to radioscopic image, realize the splicing fusion of blade defect image, pre-process and defect is carried
Take, be qualitative, positioning, quantitative analysis.
Claims (3)
1. the extraction of gas turbine blades defect and analysis method based on region segmentation detection, it is characterised in that including following step
Suddenly：
1) blade is divided into more than two regions using galvanized wire according to the size of blade, thickness Curvature varying, utilizes number
Word ray equipment (DR) carries out detecting a flaw inside subregion to gas turbine blades, and the image obtained to flaw detection is characterized by white galvanized wire
Image mosaic is carried out, obtains the blade overall intensity image A for including galvanized wire；
2) galvanized wire is moved, avoids the position where galvanized wire during subregion first, do two subzones, blade is similarly divided into synchronization
Rapid 1) equal number of region, gas turbine blades are carried out using digital radial to detect a flaw inside subregion, the figure obtained to flaw detection
As the carry out image mosaic characterized by white galvanized wire, the blade overall intensity image B for including galvanized wire is obtained；
3) by image A and image B on the basis of blade entirety common first edges feature, progress image co-registration obtains entirety and is free of galvanized wire
X ray gray scale original image；
4) X ray gray scale original image is pre-processed, obtains the X-ray detection image of blade；
5) X-ray detection image is handled to obtain the binary image of defect protrusion, bidimensional is established on binary image
Coordinate system, positioning analysis is carried out to existing internal flaw；
6) the girth sum of all pixels and area sum of all pixels of defect in binary image are calculated, actual lack is obtained by conversion
Size is fallen into, quantitative analysis is carried out to existing internal flaw, by extracting related feature --- axial ratio, girth to defect
Area ratio, is classified based on the feature extracted to defect, so as to realize the qualitative analysis to blade interior defect；
What described step 1) image A was obtained concretely comprises the following steps：
1.1) 6 flaw detections are carried out according to the exposure parameter of the Ith subregion to the Ith subregion of blade using X-ray machine, exposure parameter includes
Electric current, voltage, time for exposure, obtain the X-ray detection original image of 6 the Ith subregions；
1.2) by controlling rotatable stage that blade rotates a certain angle, control XYZ guide rails are moved X-ray machine and DDA
Detector moves so that the central point of x-ray source faces the central point of the subregion of blade the IIIth, to the IIIth subregion of blade according to
The exposure parameter of IIIth subregion carries out 6 flaw detections, obtains the X-ray detection original image of 6 the IIIth subregions；
1.3) similarly operate, obtain the V, the VI, IV, the X-ray detection original image of II subregion；
1.4) the 6 width original images obtained on each subregion, which make 6 average Removing Random Nos, to be influenceed；
1.5) characteristic point relevant to the image zooming-out white galvanized wire obtained after 1.4) handling, figure is calculated according to formula (1)
As translation rotation transformation coordinate relative between any two, image registration is realized, i.e.,：
Assuming that common ground is respectively (x between adjacent two images11,y11)、(x12,y12)、(x21,y21)、(x22,y22) by following change
Transformation parameter cos θ, sin θ, t can be tried to achieve by changing formula (1)x、ty,
<mrow>
<mfenced open = "[" close = "]">
<mtable>
<mtr>
<mtd>
<msub>
<mi>x</mi>
<mn>21</mn>
</msub>
</mtd>
<mtd>
<msub>
<mi>y</mi>
<mn>21</mn>
</msub>
</mtd>
<mtd>
<mn>1</mn>
</mtd>
<mtd>
<mn>0</mn>
</mtd>
</mtr>
<mtr>
<mtd>
<msub>
<mi>x</mi>
<mn>22</mn>
</msub>
</mtd>
<mtd>
<msub>
<mi>y</mi>
<mn>22</mn>
</msub>
</mtd>
<mtd>
<mn>1</mn>
</mtd>
<mtd>
<mn>0</mn>
</mtd>
</mtr>
<mtr>
<mtd>
<msub>
<mi>y</mi>
<mn>21</mn>
</msub>
</mtd>
<mtd>
<mrow>
<mo>-</mo>
<msub>
<mi>x</mi>
<mn>21</mn>
</msub>
</mrow>
</mtd>
<mtd>
<mn>0</mn>
</mtd>
<mtd>
<mn>1</mn>
</mtd>
</mtr>
<mtr>
<mtd>
<msub>
<mi>y</mi>
<mn>22</mn>
</msub>
</mtd>
<mtd>
<mrow>
<mo>-</mo>
<msub>
<mi>x</mi>
<mn>22</mn>
</msub>
</mrow>
</mtd>
<mtd>
<mn>0</mn>
</mtd>
<mtd>
<mn>1</mn>
</mtd>
</mtr>
</mtable>
</mfenced>
<mfenced open = "[" close = "]">
<mtable>
<mtr>
<mtd>
<mi>cos</mi>
<mi>&theta;</mi>
</mtd>
</mtr>
<mtr>
<mtd>
<mi>s</mi>
<mi>i</mi>
<mi>n</mi>
<mi>&theta;</mi>
</mtd>
</mtr>
<mtr>
<mtd>
<msub>
<mi>t</mi>
<mi>x</mi>
</msub>
</mtd>
</mtr>
<mtr>
<mtd>
<msub>
<mi>t</mi>
<mi>y</mi>
</msub>
</mtd>
</mtr>
</mtable>
</mfenced>
<mo>=</mo>
<mfenced open = "[" close = "]">
<mtable>
<mtr>
<mtd>
<msub>
<mi>x</mi>
<mn>11</mn>
</msub>
</mtd>
</mtr>
<mtr>
<mtd>
<msub>
<mi>x</mi>
<mn>12</mn>
</msub>
</mtd>
</mtr>
<mtr>
<mtd>
<msub>
<mi>y</mi>
<mn>11</mn>
</msub>
</mtd>
</mtr>
<mtr>
<mtd>
<msub>
<mi>y</mi>
<mn>12</mn>
</msub>
</mtd>
</mtr>
</mtable>
</mfenced>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
</mrow>
1.6) splicing is carried out to 6 width images based on the result after registration, image after splicing is determined according to formula (2)-formula (6)
Size：Wherein w1、h1、w2、h2For the width of two images to be spliced, height；
<mrow>
<mfenced open = "[" close = "]">
<mtable>
<mtr>
<mtd>
<mrow>
<mi>c</mi>
<mi>o</mi>
<mi>s</mi>
<mi>&theta;</mi>
</mrow>
</mtd>
<mtd>
<mrow>
<mi>s</mi>
<mi>i</mi>
<mi>n</mi>
<mi>&theta;</mi>
</mrow>
</mtd>
<mtd>
<msub>
<mi>t</mi>
<mi>x</mi>
</msub>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mo>-</mo>
<mi>s</mi>
<mi>i</mi>
<mi>n</mi>
<mi>&theta;</mi>
</mrow>
</mtd>
<mtd>
<mrow>
<mi>c</mi>
<mi>o</mi>
<mi>s</mi>
<mi>&theta;</mi>
</mrow>
</mtd>
<mtd>
<msub>
<mi>t</mi>
<mi>y</mi>
</msub>
</mtd>
</mtr>
<mtr>
<mtd>
<mn>0</mn>
</mtd>
<mtd>
<mn>0</mn>
</mtd>
<mtd>
<mn>1</mn>
</mtd>
</mtr>
</mtable>
</mfenced>
<mo>&times;</mo>
<mfenced open = "[" close = "]">
<mtable>
<mtr>
<mtd>
<mn>1</mn>
</mtd>
<mtd>
<mn>1</mn>
</mtd>
<mtd>
<msub>
<mi>w</mi>
<mn>2</mn>
</msub>
</mtd>
<mtd>
<msub>
<mi>w</mi>
<mn>2</mn>
</msub>
</mtd>
</mtr>
<mtr>
<mtd>
<mn>1</mn>
</mtd>
<mtd>
<mn>1</mn>
</mtd>
<mtd>
<msub>
<mi>h</mi>
<mn>2</mn>
</msub>
</mtd>
<mtd>
<msub>
<mi>h</mi>
<mn>2</mn>
</msub>
</mtd>
</mtr>
<mtr>
<mtd>
<mn>1</mn>
</mtd>
<mtd>
<mn>1</mn>
</mtd>
<mtd>
<mn>1</mn>
</mtd>
<mtd>
<mn>1</mn>
</mtd>
</mtr>
</mtable>
</mfenced>
<mo>=</mo>
<msub>
<mi>A</mi>
<mrow>
<mi>a</mi>
<mo>+</mo>
<mi>b</mi>
</mrow>
</msub>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>2</mn>
<mo>)</mo>
</mrow>
</mrow>
xmin=min (Aa+b(1,:)0) (3)
xmax=max (Aa+b(1,:)w1) (4)
ymin=min (Aa+b(2,:)0) (5)
ymax=max (Aa+b(2,:)h1) (6)
1.7) by after image rotation translation transformation registration to be spliced, stitching image is treated using formula (7) bilinear interpolation and entered
Row interpolation resampling, obtain preliminary spliced image；
V (x, y)=ax+by+cxy+d (7)
The unknown equation that wherein 4 coefficients a, b, c, d are write out by 4 (x, y) adjoint points determines that v (x, y) is resampling gray value；
1.8) later stage smoothing processing is carried out to spliced image, is changed using 3 × 3 template smooth grey of formula (8) and protruded
Splicing line, obtain the blade overall intensity image A for including galvanized wire；
<mrow>
<mfenced open = "[" close = "]">
<mtable>
<mtr>
<mtd>
<mn>1</mn>
</mtd>
<mtd>
<mn>1</mn>
</mtd>
<mtd>
<mn>1</mn>
</mtd>
</mtr>
<mtr>
<mtd>
<mn>1</mn>
</mtd>
<mtd>
<mn>1</mn>
</mtd>
<mtd>
<mn>1</mn>
</mtd>
</mtr>
<mtr>
<mtd>
<mn>1</mn>
</mtd>
<mtd>
<mn>1</mn>
</mtd>
<mtd>
<mn>1</mn>
</mtd>
</mtr>
</mtable>
</mfenced>
<mo>&times;</mo>
<mfrac>
<mn>1</mn>
<mn>9</mn>
</mfrac>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>8</mn>
<mo>)</mo>
</mrow>
<mo>.</mo>
</mrow>
2. the extraction of gas turbine blades defect and analysis method according to claim 1 based on region segmentation detection, its
It is characterised by：Described step 5) defect location analysis comprises the following steps that：
5.1) the X-ray detection image of integral blade is smoothed with the average value template of 3 × 3 pixels, simulated
Go out flawless image；
5.2) flawless analog image and overall X-ray detection image are made into difference operation, obtains the difference of defect protrusion
Image；Because the existing gray value of defect area is the black region near 0, there is gray value again for the white portion near 255, institute
The image that black defect area protrudes and White Defects region protrudes is obtained to need to do difference operation twice respectively；
5.3) error image protruded defect makees Otsu automatic threshold segmentation computings and obtains containing defective binary image；
5.4) binary image that black defect area protrudes and White Defects region protrudes is synthesized into computing, and does region life
It is long, obtain the binaryzation defect image of the complete fidelity in region；
5.5) rectangular coordinate system as the origin of coordinates, is established in the upper left corner in gas turbine blades region using on image；
5.6) transverse and longitudinal coordinate of each defect is tried to achieve with the coordinate system of foundation；
5.7) the abscissa X by 5.6) obtainingn, ordinate YnThe barycentric coodinates of each defect are calculated, realizes and defect is determined
Position analysis, center of gravity calculation formula are as follows；
<mrow>
<mtable>
<mtr>
<mtd>
<mrow>
<mover>
<mi>x</mi>
<mo>&OverBar;</mo>
</mover>
<mo>=</mo>
<mfrac>
<mrow>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>x</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>M</mi>
</munderover>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>y</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>N</mi>
</munderover>
<mi>x</mi>
<mi>f</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>x</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>M</mi>
</munderover>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>y</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>N</mi>
</munderover>
<mi>f</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
</mrow>
</mfrac>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mover>
<mi>y</mi>
<mo>&OverBar;</mo>
</mover>
<mo>=</mo>
<mfrac>
<mrow>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>x</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>M</mi>
</munderover>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>y</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>N</mi>
</munderover>
<mi>y</mi>
<mi>f</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>x</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>M</mi>
</munderover>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>y</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>N</mi>
</munderover>
<mi>f</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>,</mo>
<mi>y</mi>
<mo>)</mo>
</mrow>
</mrow>
</mfrac>
</mrow>
</mtd>
</mtr>
</mtable>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>12</mn>
<mo>)</mo>
</mrow>
<mo>.</mo>
</mrow>
3. the extraction of gas turbine blades defect and analysis method according to claim 2 based on region segmentation detection, its
It is characterised by：Described step 6) Flaw discrimination analysis comprises the following steps that：
6.1) the X-ray detection image of integral blade is subjected to rim detection using single order edge detection operator；
6.2) opening operation is made to the image after rim detection, eliminates the influence of noise spot；
6.3) image after step 6.2) processing is eliminated with doing add operation containing defective bianry image in step 5.3)
The template smoothly influence to defect image；
6.4) image obtained to step 6.3) detects with a second order edge detective operators to it；
6.5) image that step 6.4) obtains is calculated, obtains the Euclidean distance between pixel two-by-two, find out it is therein most
Big value is the most major axis of the defect image, and calculates the pixel transverse and longitudinal coordinate at most major axis both ends；
6.6) slope K is tried to achieve by the transverse and longitudinal coordinate obtained in step 6.5)；
6.7) calculation procedure 6.4) in image straight slope is the pixel set of (- 1/K) between pixel two-by-two；
6.8) pixel set that step 6.7) obtains is calculated, obtains corresponding Euclidean distance, find out maximum therein,
The as minor axis length of the defect；
6.9) the ratio between major axis and short axle of defect and girth and area ratio are calculated；
6.10) by obtained major and minor axis length ratio and girth area ratio, the common type for determining defect, i.e. length axial length
When the ratio between degree, girth area ratio are all higher than the threshold value each given, defect type is crack defect, is otherwise lacked for shrinkage cavity class
Fall into, be so achieved that the qualitative analysis to defect.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510068469.3A CN104730091B (en) | 2015-02-10 | 2015-02-10 | The extraction of gas turbine blades defect and analysis method based on region segmentation detection |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510068469.3A CN104730091B (en) | 2015-02-10 | 2015-02-10 | The extraction of gas turbine blades defect and analysis method based on region segmentation detection |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104730091A CN104730091A (en) | 2015-06-24 |
| CN104730091B true CN104730091B (en) | 2018-01-16 |
Family
ID=53454190
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510068469.3A Expired - Fee Related CN104730091B (en) | 2015-02-10 | 2015-02-10 | The extraction of gas turbine blades defect and analysis method based on region segmentation detection |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104730091B (en) |
Families Citing this family (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106932416B (en) * | 2017-02-16 | 2019-10-11 | 西安交通大学 | Gas turbine blades internal flaw three-dimensional parameter extracting method based on digital radial |
| CN106845011B (en) * | 2017-02-16 | 2020-03-31 | 西安交通大学 | Large-scale gas turbine blade digital ray partition detection method |
| CN107052320A (en) * | 2017-04-12 | 2017-08-18 | 沈阳理工大学 | Auto-pouring System and sprue cup liquid level recognition methods based on image detection |
| CN107255673A (en) * | 2017-05-05 | 2017-10-17 | 西安交通大学 | High temperature blade internal flaw three dimensional lossless detection method based on ultrasonic phase array |
| CN108506170A (en) * | 2018-03-08 | 2018-09-07 | 上海扩博智能技术有限公司 | Fan blade detection method, system, equipment and storage medium |
| CN108229080B (en) * | 2018-03-26 | 2020-06-19 | 西安交通大学 | Optimization method for model unknown gas turbine blade digital ray subarea transillumination direction |
| CN108986071B (en) * | 2018-05-31 | 2020-12-01 | 上海扩博智能技术有限公司 | Automatic blade detection and tracking method, system, equipment and storage medium |
| CN108956655B (en) * | 2018-07-06 | 2019-12-20 | 西安交通大学 | Digital ray detection partitioning method for high-temperature blades of gas turbine with unknown model |
| CN108872277A (en) * | 2018-07-23 | 2018-11-23 | 北方夜视技术股份有限公司 | X-ray non-destructive detection device based on lobster eye lens |
| CN109459451A (en) * | 2018-12-13 | 2019-03-12 | 中国航空工业集团公司上海航空测控技术研究所 | A kind of metal inside testing of small cracks method based on ray contrast |
| CN112858351A (en) * | 2019-11-28 | 2021-05-28 | 北京金自天正智能控制股份有限公司 | X-ray nondestructive inspection method based on machine vision |
| CN111583216A (en) * | 2020-04-30 | 2020-08-25 | 深圳比特微电子科技有限公司 | Defect detection method for PCBA |
| CN112880588A (en) * | 2021-01-29 | 2021-06-01 | 广州信邦智能装备股份有限公司 | Blade three-dimensional data acquisition method based on combination of segmented scanning and splicing fusion |
| CN113237908B (en) * | 2021-04-30 | 2022-10-11 | 北京科技大学 | Method for evaluating crack sensitivity of hypo-peritectic steel |
| CN113363024B (en) * | 2021-06-02 | 2021-11-16 | 广州市广惠通线缆有限公司 | Cable foaming layer nitrogen filling control method and system |
| CN113670958B (en) * | 2021-09-02 | 2022-12-06 | 西安交通大学 | Gas turbine blade defect identification method based on X-ray attenuation coefficient difference |
| CN114354623A (en) * | 2021-12-30 | 2022-04-15 | 苏州凌云视界智能设备有限责任公司 | Weak mark extraction algorithm, device, equipment and medium |
Family Cites Families (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPS63120244A (en) * | 1986-09-19 | 1988-05-24 | ゼネラル・エレクトリック・カンパニイ | Device for inspecting body and method of improving signal-to-noise ratio |
| JP2002082199A (en) * | 2000-09-08 | 2002-03-22 | Shimadzu Corp | Radiation inspection system |
| CN1480301A (en) * | 2003-04-30 | 2004-03-10 | 哈尔滨工业大学 | Method for synchronous tracking pipeline robot through X ray detecting realtime image |
| GB2403799B (en) * | 2003-07-11 | 2006-04-12 | Rolls Royce Plc | Image-based measurement |
| CN101126725B (en) * | 2007-09-24 | 2010-12-15 | 舒嘉 | Method for realizing image reconstruction by adopting X ray dimension photograph |
| CN101435784B (en) * | 2008-10-14 | 2012-01-18 | 重庆大学 | Turbine blade CT detection device and detection method |
| JP2010253194A (en) * | 2009-04-28 | 2010-11-11 | Fujifilm Corp | Radiation phase imaging apparatus |
| EP2282050B1 (en) * | 2009-06-30 | 2016-03-16 | Siemens Aktiengesellschaft | Method to inspect a wind turbine blade |
| FR2960642B1 (en) * | 2010-05-28 | 2012-07-13 | Snecma | NON-DESTRUCTIVE CONTROL METHOD AND DEVICE FOR IMPLEMENTING THE METHOD |
-
2015
- 2015-02-10 CN CN201510068469.3A patent/CN104730091B/en not_active Expired - Fee Related
Also Published As
| Publication number | Publication date |
|---|---|
| CN104730091A (en) | 2015-06-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104730091B (en) | The extraction of gas turbine blades defect and analysis method based on region segmentation detection | |
| CN113362326B (en) | Method and device for detecting defects of welding spots of battery | |
| CN102636490B (en) | Method for detecting surface defects of dustproof cover of bearing based on machine vision | |
| CN105547602B (en) | A kind of telemeasurement method of subway tunnel section of jurisdiction percolating water | |
| CN107742286A (en) | A kind of polysilicon solar battery slice EL tests crack defect detection method | |
| CN105046697B (en) | A kind of beer bottleneck localization method assessed based on multiple random loop truss and degree of fitting | |
| CN102974551A (en) | Machine vision-based method for detecting and sorting polycrystalline silicon solar energy | |
| CN114120317B (en) | Optical element surface damage identification method based on deep learning and image processing | |
| CN110108715B (en) | Defect detection method for transparent parallel flat plate | |
| CN206223679U (en) | Photomask detection device | |
| NL2026437A (en) | Deep convolutional neural network based high-throughput method for determining alkali spreading value of rice grain | |
| CN108388713A (en) | Gas turbine blades search coverage adaptivenon-uniform sampling and exposure parameter optimization method | |
| CN115330802B (en) | Method for extracting debonding defect of X-ray image of carbon fiber composite gas cylinder | |
| CN110097547A (en) | A kind of automatic testing method that the weld seam egative film based on deep learning is faked | |
| CN113554587A (en) | Molten pool image geometric feature extraction method and system based on deep learning | |
| CN115601359A (en) | Welding seam detection method and device | |
| CN107145896A (en) | Dysnusia identifying system based on fluorescentmagnetic particle(powder) | |
| CN106596562A (en) | Magnet ring appearance quality on-line detection apparatus based on machine visual sense and method thereof | |
| CN109544513A (en) | A kind of steel pipe end surface defect extraction knowledge method for distinguishing | |
| WO2012122542A2 (en) | Intelligent airfoil component surface imaging inspection | |
| CN110426395B (en) | Method and device for detecting surface of solar EL battery silicon wafer | |
| CN103177426B (en) | Morphology-based strong-interference laser edge image restoration method | |
| CN106770333A (en) | A kind of BGA defects automatic testing method | |
| CN106814074B (en) | A kind of cladding nuclear fuels surface defects detection system and method | |
| CN112183593A (en) | Road damage detection method based on DSSD algorithm |
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: 20180116 Termination date: 20210210 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |