CN107870181A - A kind of later stage recognition methods of composite debonding defect - Google Patents
A kind of later stage recognition methods of composite debonding defect Download PDFInfo
- Publication number
- CN107870181A CN107870181A CN201710469654.2A CN201710469654A CN107870181A CN 107870181 A CN107870181 A CN 107870181A CN 201710469654 A CN201710469654 A CN 201710469654A CN 107870181 A CN107870181 A CN 107870181A
- Authority
- CN
- China
- Prior art keywords
- defect
- image
- pixel
- region
- composite
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N25/00—Investigating or analyzing materials by the use of thermal means
- G01N25/72—Investigating presence of flaws
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10048—Infrared image
Abstract
The present invention discloses a kind of later stage recognition methods of composite debonding defect, pass through stroboscopic optical light pulses infrared thermal imaging, obtain the response sequence that the surface temperature containing defect information changes over time, vectorization is per frame thermal image, thermal image signal reconstruct is carried out again, 6 obtained frame logarithm multinomial coefficient images are compressed, defect information can be strengthened；Select the best image of an effect frame, according to image deflects characteristic, divided the image into using region-growing method as different zones, initial growth seed point is more optionally used as in non-defective region, then the certain growth criterion of basis finally realizes defect Segmentation by the potting gum for having same nature in sub-pixel neighborhood with seed to sub-pixel region.The inventive method can preferably solve low resolution problem in flash lamp infrared thermal imaging method defects detection, the contrast being remarkably improved between defect and non-defective region.
Description
Technical field
The invention belongs to Non-Destructive Testing and assessment technology field, specifically, relates generally to compound for fibre reinforced
The detection sensitivity of material debonding defect.
Background technology
Carbon fibre reinforced composite (carbon fiber reinforced plastic CFRP) is due to than strong
The series of advantages such as degree is high, specific modulus is big, anti-fatigue performance is good, the coefficient of expansion is low and fail-safety is good, have become boat
The indispensable important feature material of the high-technology fields such as empty space flight.As composite is in wind power generation and aerospace component
The continuous enhancing of upper application percentage, to ensure safety of being on active service, the internal soundness for monitoring composite structure is widely paid close attention to.
Therefore, all kinds of non-destructive testing technologies (Nondestructive testing and evaluation NDTE) are also more and more
In the overall process for being molded, assemble, test, safeguarding and using applied to composite structure.Stroboscopic optical light pulses infrared thermal imaging
(Flash Pulsed Thermography FPT) detection technique utilizes object heat radiation caused by structure or material difference
Characteristic is different, uses active heating means to heat test specimen to excite display surface and hide various under surface
Defect and damage, continuously recorded using thermal imaging system and reflect test specimen each point in the form of gray scale difference or pcolor on a display screen
Temperature change, then gathered thermal image sequence is analyzed by heat waves and computer image processing technology and obtains defect
The relevant informations such as shape, depth, property, so as to realize the purpose of detection.This technical security is pollution-free, and large area, nothing can be achieved
The quick detection of contact.
Because thermal-induced imagery has, resolution ratio is relatively low, blurred edges, the characteristics of mixing a variety of noises, in order to improve defect
The accuracy and efficiency of detection, many data processing algorithms are used for the extraction of image information, such as independent component analysis
(Independent Component Analysis ICA), principal component analysis (Principal Component Analysis
PCA), thermal image signal reconstruct (Thermographic Signal Reconstruction TSR), impulse phase thermal imaging
(Pulsed Phase Thermography PPT) etc..But in actual applications, a kind of method income effect of single use is simultaneously
It is non-very good, so as to use Multi-category efficient combination.
The content of the invention
The present invention is to be based on heat waves and object radiation characteristic, there is provided a kind of stroboscopic optical light pulses infrared thermal imaging is fine to carbon
The automatic detection identification of dimension enhancing composite debonding defect, with the realization sudden strain of a muscle that directly processing is recorded for infrared thermography
The thermal map video of light lamp pulse thermal imaging, can be to the separation and judgement of debonding defect.
To realize foregoing invention, a kind of later stage recognition methods of composite debonding defect of the present invention, comprise the following steps：
S1 obtains data：
S11 encourages composite debonding defect by light pulse THERMAL IMAGING NONDESTRUCTIVE TESTING system with Halogen lamp LED Par64
Sample, changed by FLIR thermal infrared imager collecting samples surface temperature, obtain the surface temperature containing defect information with the time
The response video of change；
S12 chooses arrangement to each frame thermal image by leu time, and vectorization is per frame thermal image one new matrix of framework；
S2 is pre-processed:
S21 carries out thermal image signal reconstruct to new matrix, and compression obtains 6 frame logarithm multinomial coefficient images；
S22 chooses a most clearly two field picture in the image after overcompression；
S3 defect recognitions and segmentation:
S31 is divided the image into as different zones to the imagery exploitation region-growing method of selection；
S32 selects non-defective high luminance pixel point as seed point to the region of each segmentation；
S33 judges pixel around seed point whether within the scope of defined growing threshold, by satisfactory pixel
Point is merged into sub-pixel region, and undesirable pixel then negates, and is merged the new pixel of gained and is continued conduct
Seed grows to surrounding, then terminates until all pixels point is all treated, and output image, realizes defect Segmentation.
The similarity criterion of region growing is in the step S33：Based on interregional gray scale difference, average gray it is uniform
Measure.
The it is proposed of the recognition methods of composite debonding defect of the present invention, it is mainly based upon the thing of composite debonding defect
Manage characteristic, under the excitation of Halogen lamp LED, the heterogeneity phantom of spectrum in the composite by material anisotropic influence and
Difference, when material existing defects (such as unsticking), hot-fluid, which propagates conduct, can be forced to shift, so as to form the hot-fluid of different temperatures point
Cloth, collection hot-zone can be formed in the region of unsticking, and other regional temperatures are then relatively low, material surface heat distribution is thermal image sequence
Row.The blending algorithm based on region growing and thermal imaging signal reconstruct, it is of the invention in thermal characteristics information extraction process
Core.The inventive method can preferably solve low resolution problem in flash lamp infrared thermal imaging method defects detection, can be notable
Improve the contrast between defect and non-defective region.
Brief description of the drawings
Fig. 1 is composite debonding defect detection of the present invention and identification process figure；
(Fig. 2 .1 the i-th two field picture vectorizations, Fig. 2 .2 image sequences are converted into square to Fig. 2 image sequence vectorizations schematic diagram
Battle array)；
Fig. 3 is second order heat conduction imaging signal restructing algorithm schematic diagram；
Fig. 4 is region-growing method growing strategy flow chart；
Fig. 5 is Data Processing in Experiment design sketch.
Embodiment
The embodiment of invention is described below in conjunction with the accompanying drawings, it is more preferable in order to the technical staff of this neighborhood
Understand the present invention.Requiring particular attention is that in the following description, when known function and the detailed description of design perhaps
When can desalinate the main contents of the present invention, these descriptions will be ignored herein.
Fig. 1 is composite debonding defect detection of the present invention and the flow chart of identification, mainly includes the light containing defect sample
The acquisition of pulse heat imaging video sequence, data dyad is read, pretreatment and image based on thermal imaging signal reconstruction
Selection, and identify the defects of using region-growing method as core and partitioning portion, comprise the following steps：
1st, data are obtained
First, by light pulse THERMAL IMAGING NONDESTRUCTIVE TESTING system, lacked with Halogen lamp LED Par64 excitations composite unsticking
Sample is fallen into, is changed by FLIR thermal infrared imager collecting samples surface temperature, obtains the table containing defect information on computers
The response video that face temperature changes over time.First to the frame thermal image of gained FPT thermal maps video one by leu time value, such as Fig. 2 .1
It is shown, and ordered arrangement, obtain column vector vec (Y ' (i)).For another example shown in Fig. 2 .2, by image sequence successively vectorization and structure
New matrix is put up, each frame is converted into column vector, and whole video sequence, which merges, is arranged to make up new matrix.
2nd, pre-process
Thermal image signal reconstruct (TSR) compression as shown in Figure 3, i.e. logarithm multinomial are carried out to new matrix obtained by vectorization
Fitting obtains high-definition picture, to eliminate the even influence of uneven illumination.Thermal imaging signal reconstruction belongs to prior art, is a kind of
The treatment technology of the room and time resolution ratio of thermal image sequence is improved, concrete principle and derivation of equation step can be in papers
《Balageas D L,Roche J M,Leroy F H,et al.The thermographic signal
reconstruction method:A powerful tool for the enhancement of transient
thermographic images[J].Biocybernetics and Biomedical Engineering,2015,35(1):
1-9.》In obtain.TSR assumes to show as being evenly applied to the ideal on the surface of semi-infinite body for the temperature of area free from defect
The attenuation curve that the dimension solutions of the Fourier diffusion equation of pulse provide, test specimen table can only be got in actually detected middle thermal imaging system
Face temperature, so the temperature change of surface of test piece can be expressed as：
WhereinIt is thermal diffusivity, Q is the energy of surface irradiation, and t is the time.It is more accurately warm for thermal imaging system
Spending response can be in the hope of its surface temperature difference Δ Tsurface(t), this parameter can evaluate the observability of defect, and enhancing picture quality is simultaneously
Calculate defect parameters.
TSR technologies are in given thermal image sequence, and all pixels point changes with time rule, wherein time response
Curve equation (2) may switch to log-domain, and linear relationship can more intuitively observe useful information.
Equation (3) shows, regardless of the hot property for being detected material, for preferable non-defective region, logarithmic decrement
Response by be all the time slope be -1/2 straight line.Actually due to a variety of causes, for example, background radiation, the response of non-linear camera with
And defect area, logarithmic data will deviate from this ideal relationship of above formula.Logarithm Temperature Evolution at each pixel can be n times by following formula
Polynomial function carrys out approximate fits：
ln(ΔTsurface(t))=a0+a1ln(t)+a2[ln(t)]2···+an[ln(t)]n (4)
This approximation eliminates high frequency time noise and by the multinomial coefficient of thermography sequence compaction to n+1 frames, etc.
The logarithmic curve that formula (4) creates, include the difference of defect and non-defective region thermal response on a timeline.Counted using equation (4)
The first differential and second-order differential to ln (t) are calculated, its single order leads to obtain its differential temperature Tsurface(t) change on a timeline
Rate, and second order is led and represents skin temperature profile concavity and convexity on a timeline, the reaction defect area that can be become apparent from and non-
The response difference of the temperature change of defect area pixel, strengthens defect information.In addition, 5 ranks or 6 rank multinomials can effectively fill
When low pass filter is suitable for anisotropic fibre reinforced modeling with smoothed data without rebuilding noise, 7 ranks to 9 rank multinomials
Material and cellular sandwich.
In the image after TSR compressions, a most clearly two field picture is chosen.
3rd, defect recognition and segmentation
To choosing image, Region growing segmentation is carried out.Segmentation is the important step of graphical analysis and processing, general by set
Read and image segmentation is defined as below：Order set R represents whole image region, and the segmentation to R can be regarded as is divided into N number of satisfaction by R
The nonvoid subset R of five conditions once1,R2···RN：
(1)
(2) to all i and j, i ≠ j, there is Ri∩Rj≠φ；
(3) to i=1,2, N, there are P (Ri)=true；
(4) to i ≠ j, there are P (Ri∪Rj)=false；
(5) to i=1,2, N, RiIt is the region of connection.
Wherein P (Ri) to all in set RiThe logical predicate of middle element, φ represent empty set.
Most of image segmentation algorithms are in the features, such as image such as similitude or discontinuity based on image pixel
Point, edge and line between different pixels.Wherein, the basic thought of region-growing method used herein is by with similar quality
Pixel set get up to form region, its operating process to the picture of selection as shown in figure 4, carry out gray processing processing, then to every
Starting point of the seed point as growth is found out in the individual region for needing to split, and then judges whether the pixel around seed point is advising
Within the scope of fixed growing threshold, satisfactory pixel is merged into sub-pixel region, it is undesirable
Pixel then negates, and merges the new pixel of gained and continues to grow to surrounding as seed, until all pixels point all treats then
Terminate, and output image is end product.
In the practical application of the present invention, the process chart of corresponding upper zone growth method is, it is necessary to solve three keys
Problem：(1) correct sub-pixel is selected：Non-defective high luminance pixel point is selected as seed point for this categorical data；(2)
Determining in growth course can be by growth criterion that adjacent pixel is included：That is threshold range, select herein based on interregional
Gray scale difference, the uniform measure of average gray can be as the similarity criterion of region growing, if certain image-region R, its
Middle pixel count is N, then average is expressed as
Wherein m is average, and region R uniform measure is：
max|f(x,y)-m|(x,y)∈0＜ K (7)
Wherein K is threshold value, and above formula represents that in the R of region the difference of each grey scale pixel value and average is no more than a certain threshold k
When, its uniform measure be it is true, will this pixel be included into seed point.It can be obtained by the repetition experiment of multi-group data, pin
(accompanying drawing 5.4 is acquired results when threshold value is 1.5, and threshold value arrives for 1.3 between being 1.3 to 1.8 to this type flaw threshold range
Result is similar to 1.5 when 1.8)；(3) stop condition of growth is determined：After all pixels point is differentiated, there is no full
When foot adds the pixel of seed region, region growing stops.
The existing following region-growing method that illustrates is regular, carries out growth segmentation to a matrix of areas A, it is desirable to be partitioned into
Gray value is 4 to 6 region, if 5 be seed, the exhausted of gray scale difference is taken with seed to the value of each pixel around since seed
To value, as thresholding T=1, then growth result is A1, it is seen that the pixel that the gray value around seed is 4,5,6 is all by well
Wrap among growth district, and having arrived the pixel that boundary gray value is 0,1,2,7 all becomes border, though the 5 of the upper right corner
It can also so turn into seed, but because without the pixel for meeting thresholding, then the value around it does not change around it；When taking thresholding
During T=3, the pixel that gray value is 2 to 8 is all included into growth district, the growth result obtained newly is A2。
The one group of composite material test piece for containing debonding defect experimental data is selected, wherein Fig. 5 .1 are sample defect maps,
It can thus be appreciated that the size of debonding defect, depth and position, Fig. 5 .2 are the more excellent detection effects chosen from untreated image sequence
Fruit is schemed, and as seen from the figure, least a portion of defect can be detected clearly, but most defect combines together with background, in figure
Circular mark diameter is less than the defects of 10mm Detection results unobvious, without the obvious temperature difference compared with non-defective region.Fig. 5 .3 are
It is artificial after second order leads TSR pretreatments to choose the preferable image of effect, the temperature change of surface of test piece can be allowed more uniform,
The defects of a diameter of 10mm, 6mm, is high-visible, improves contrast and resolution ratio.Fig. 5 .4 are the pretreatment image warps to selection
Above-mentioned zone growth method (RGM) handles acquired results, also can be clear at circles mark at difficult detection, the defects of minimum 3mm
It is discernable.
By the analysis to result, defect area thresholding is 0 in final gained bianry image, and non-defective area thresholding is 1, the spy
Sign extraction algorithm can strengthen the contrast of defect area and non-defective region, improve resolution ratio, will be combined together originally with background
The defects of information extract.
Claims (2)
1. a kind of later stage recognition methods of composite debonding defect, it is characterised in that step is：
S1 obtains data：By light pulse THERMAL IMAGING NONDESTRUCTIVE TESTING system, composite unsticking is encouraged with Halogen lamp LED Par64
Defect sample, by FLIR thermal infrared imager collecting samples surface temperature change, obtain the surface temperature containing defect information with
The response video of time change；
S2 is pre-processed:Each frame thermal image is chosen by leu time and arranged, the every frame thermal image one new matrix of framework of vectorization,
Thermal image signal reconstruct is carried out to new matrix again, compression obtains 6 frame logarithm multinomial coefficient images；
S3 defect recognitions and segmentation:
S31 chooses a most clearly two field picture in the image after overcompression, using region-growing method divide the image into for
Different zones；
S32 selects non-defective high luminance pixel point as seed point to the region of each segmentation；
S33 judges that whether within the scope of defined growing threshold, satisfactory pixel is closed for pixel around seed point
And into sub-pixel region, undesirable pixel then negates, merge the new pixel of gained and continue as seed
Grow to surrounding, then terminate until all pixels point is all treated, and output image, realize defect Segmentation.
A kind of 2. later stage recognition methods of composite debonding defect according to claim 1, it is characterised in that the step
The similarity criterion of region growing is in rapid S33：Based on interregional gray scale difference, the uniform measure of average gray.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710469654.2A CN107870181A (en) | 2017-06-20 | 2017-06-20 | A kind of later stage recognition methods of composite debonding defect |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710469654.2A CN107870181A (en) | 2017-06-20 | 2017-06-20 | A kind of later stage recognition methods of composite debonding defect |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107870181A true CN107870181A (en) | 2018-04-03 |
Family
ID=61761852
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710469654.2A Pending CN107870181A (en) | 2017-06-20 | 2017-06-20 | A kind of later stage recognition methods of composite debonding defect |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107870181A (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108548846A (en) * | 2018-06-21 | 2018-09-18 | 电子科技大学 | Bearing device subsurface defect extracting method based on thermal image AP clusters |
| CN110376249A (en) * | 2019-07-02 | 2019-10-25 | 福建师范大学福清分校 | A kind of current vortex thermal imaging testing method and device |
| CN110806427A (en) * | 2019-11-27 | 2020-02-18 | 云南电网有限责任公司电力科学研究院 | Online detection method and system for internal defects of circuit composite insulator |
| CN111429367A (en) * | 2020-03-06 | 2020-07-17 | 暨南大学 | Thermal image data processing method for composite material layered damage detection |
| CN111537075A (en) * | 2020-04-14 | 2020-08-14 | 重庆中科云从科技有限公司 | Temperature extraction method, device, machine readable medium and equipment |
| CN112489016A (en) * | 2020-11-30 | 2021-03-12 | 浙江工业大学 | Local sensitivity discriminant analysis method for composite material defect thermal image |
| CN112837294A (en) * | 2021-02-07 | 2021-05-25 | 浙江工业大学 | Thermal imaging defect detection method based on convolution self-encoder image amplification |
| CN113362306A (en) * | 2021-06-07 | 2021-09-07 | 中山大学 | Packaged chip defect detection method based on deep learning |
| CN115330802A (en) * | 2022-10-17 | 2022-11-11 | 山东大学 | Carbon fiber composite material gas cylinder X-ray image debonding defect extraction method |
| CN115439481A (en) * | 2022-11-09 | 2022-12-06 | 青岛平电锅炉辅机有限公司 | Deaerator welding quality detection method based on image processing |
| CN115861317A (en) * | 2023-02-27 | 2023-03-28 | 深圳市伟利达精密塑胶模具有限公司 | Plastic mold production defect detection method based on machine vision |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070126730A1 (en) * | 2003-11-12 | 2007-06-07 | Taiga Goto | Image processing method, image processing device, computer aided detection, and method for filtering along the time axis |
| CN101339659A (en) * | 2008-08-22 | 2009-01-07 | 北京矿冶研究总院 | Region growing image segmentation method based on rules |
| CN102590218A (en) * | 2012-01-16 | 2012-07-18 | 安徽中科智能高技术有限责任公司 | Device and method for detecting micro defects on bright and clean surface of metal part based on machine vision |
| CN102692429A (en) * | 2011-03-24 | 2012-09-26 | 中国科学院沈阳自动化研究所 | Method for automatic identification and detection of defect in composite material |
| CN103413333A (en) * | 2013-08-08 | 2013-11-27 | 宁波成电泰克电子信息技术发展有限公司 | Method for detecting concave-convex defects at tail end of alkaline battery |
| CN103487729A (en) * | 2013-09-06 | 2014-01-01 | 广东电网公司电力科学研究院 | Electrical equipment defect detection method based on fusion of ultraviolet video and infrared video |
| CN103592333A (en) * | 2013-11-13 | 2014-02-19 | 电子科技大学 | Automatic defect detection and identification method for ECPT (eddy current pulsed thermography) |
| CN104036484A (en) * | 2013-03-06 | 2014-09-10 | 株式会社东芝 | Image segmentation device, image segmentation method and medical image equipment |
| CN106468677A (en) * | 2016-09-06 | 2017-03-01 | 电子科技大学 | The automatic division method of defect in a kind of impulse eddy current thermal imaging |
-
2017
- 2017-06-20 CN CN201710469654.2A patent/CN107870181A/en active Pending
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070126730A1 (en) * | 2003-11-12 | 2007-06-07 | Taiga Goto | Image processing method, image processing device, computer aided detection, and method for filtering along the time axis |
| CN101339659A (en) * | 2008-08-22 | 2009-01-07 | 北京矿冶研究总院 | Region growing image segmentation method based on rules |
| CN102692429A (en) * | 2011-03-24 | 2012-09-26 | 中国科学院沈阳自动化研究所 | Method for automatic identification and detection of defect in composite material |
| CN102590218A (en) * | 2012-01-16 | 2012-07-18 | 安徽中科智能高技术有限责任公司 | Device and method for detecting micro defects on bright and clean surface of metal part based on machine vision |
| CN104036484A (en) * | 2013-03-06 | 2014-09-10 | 株式会社东芝 | Image segmentation device, image segmentation method and medical image equipment |
| CN103413333A (en) * | 2013-08-08 | 2013-11-27 | 宁波成电泰克电子信息技术发展有限公司 | Method for detecting concave-convex defects at tail end of alkaline battery |
| CN103487729A (en) * | 2013-09-06 | 2014-01-01 | 广东电网公司电力科学研究院 | Electrical equipment defect detection method based on fusion of ultraviolet video and infrared video |
| CN103592333A (en) * | 2013-11-13 | 2014-02-19 | 电子科技大学 | Automatic defect detection and identification method for ECPT (eddy current pulsed thermography) |
| CN106468677A (en) * | 2016-09-06 | 2017-03-01 | 电子科技大学 | The automatic division method of defect in a kind of impulse eddy current thermal imaging |
Non-Patent Citations (2)
| Title |
|---|
| 初铭强等: "《飞机材料与结构检测技术》", 31 December 2016, 上海交通大学出版社 * |
| 张金玉等: "《红外热波检测及其图像序列处理技术》", 30 June 2015, 国防工业出版社 * |
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108548846A (en) * | 2018-06-21 | 2018-09-18 | 电子科技大学 | Bearing device subsurface defect extracting method based on thermal image AP clusters |
| CN110376249A (en) * | 2019-07-02 | 2019-10-25 | 福建师范大学福清分校 | A kind of current vortex thermal imaging testing method and device |
| CN110806427A (en) * | 2019-11-27 | 2020-02-18 | 云南电网有限责任公司电力科学研究院 | Online detection method and system for internal defects of circuit composite insulator |
| CN111429367B (en) * | 2020-03-06 | 2023-09-19 | 暨南大学 | Thermal image data processing method for composite material layering damage detection |
| CN111429367A (en) * | 2020-03-06 | 2020-07-17 | 暨南大学 | Thermal image data processing method for composite material layered damage detection |
| CN111537075A (en) * | 2020-04-14 | 2020-08-14 | 重庆中科云从科技有限公司 | Temperature extraction method, device, machine readable medium and equipment |
| CN112489016A (en) * | 2020-11-30 | 2021-03-12 | 浙江工业大学 | Local sensitivity discriminant analysis method for composite material defect thermal image |
| CN112489016B (en) * | 2020-11-30 | 2024-02-20 | 浙江工业大学 | Local sensitivity discriminant analysis method for composite material defect thermal image graph |
| CN112837294A (en) * | 2021-02-07 | 2021-05-25 | 浙江工业大学 | Thermal imaging defect detection method based on convolution self-encoder image amplification |
| CN112837294B (en) * | 2021-02-07 | 2024-02-27 | 浙江工业大学 | Thermal imaging defect detection method based on convolution self-encoder image amplification |
| CN113362306A (en) * | 2021-06-07 | 2021-09-07 | 中山大学 | Packaged chip defect detection method based on deep learning |
| CN115330802A (en) * | 2022-10-17 | 2022-11-11 | 山东大学 | Carbon fiber composite material gas cylinder X-ray image debonding defect extraction method |
| CN115330802B (en) * | 2022-10-17 | 2024-01-19 | 山东大学 | Method for extracting debonding defect of X-ray image of carbon fiber composite gas cylinder |
| CN115439481B (en) * | 2022-11-09 | 2023-02-21 | 青岛平电锅炉辅机有限公司 | Deaerator welding quality detection method based on image processing |
| CN115439481A (en) * | 2022-11-09 | 2022-12-06 | 青岛平电锅炉辅机有限公司 | Deaerator welding quality detection method based on image processing |
| CN115861317B (en) * | 2023-02-27 | 2023-04-28 | 深圳市伟利达精密塑胶模具有限公司 | Plastic mold production defect detection method based on machine vision |
| CN115861317A (en) * | 2023-02-27 | 2023-03-28 | 深圳市伟利达精密塑胶模具有限公司 | Plastic mold production defect detection method based on machine vision |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107870181A (en) | A kind of later stage recognition methods of composite debonding defect | |
| CN1282132C (en) | Infra-red and visible light dynamic image interfusion method based on moving target detection | |
| CN107356599B (en) | Terahertz nondestructive testing method for ceramic matrix composite material | |
| CN104359944A (en) | Non-destructive detection method of pulse-excited infrared thermal wave phase of fixed viewing field | |
| CN106886797B (en) | The high resolution detection and recognition methods of a kind of pair of composite material debonding defect | |
| CN108550140B (en) | Confidence consistency enhancement quality evaluation method facing image set | |
| CN112489016A (en) | Local sensitivity discriminant analysis method for composite material defect thermal image | |
| Florez-Ospina et al. | From local to global analysis of defect detectability in infrared non-destructive testing | |
| CN113592794A (en) | Spine image segmentation method of 2D convolutional neural network based on mixed attention mechanism | |
| Tang et al. | Green fluorescent protein and phase contrast image fusion via dual attention residual network | |
| Zaouali et al. | Fabric wrinkling evaluation: a method developed using digital image analysis | |
| Fleuret et al. | Application of blind image quality assessment metrics to pulsed thermography | |
| CN101478694A (en) | Free view-point image quality evaluation method based on light space | |
| CN116883893A (en) | Tunnel face underground water intelligent identification method and system based on infrared thermal imaging | |
| Desiani et al. | A robust techniques of enhancement and segmentation blood vessels in retinal image using deep learning | |
| CN113763368B (en) | Multi-type damage detection characteristic analysis method for large-size test piece | |
| Liang et al. | Application of tensor decomposition methods in eddy current pulsed thermography sequences processing | |
| CN115049621A (en) | Micropipe defect detection method, device, equipment, storage medium and program product | |
| CN109886930B (en) | Thermal image defect feature extraction method based on change rate and temperature difference | |
| Li et al. | Stability of MRI radiomic features of the hippocampus: An integrated analysis of test-retest variability | |
| CN114882016B (en) | Method and system for identifying concrete defect area based on infrared temperature field time sequence information | |
| Chen et al. | Technology for multi-focus image fusion based on wavelet transform | |
| IT201800005163A1 (en) | TUMOR MASS DETECTION SYSTEM BASED ON MAGNETIC RESONANCE FOR IMAGES | |
| Wang et al. | Visual defect recognition and location for pulsed thermography images based on defect-background contrast analysis | |
| Saquib et al. | Reconstruction of Full Fiber Orientation Distribution in Molded Composites Using Deep Learning |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180403 |
|
| RJ01 | Rejection of invention patent application after publication |