CN108074235A - Carbon fiber surface defect degree method of estimation based on region growing algorithm - Google Patents
Carbon fiber surface defect degree method of estimation based on region growing algorithm Download PDFInfo
- Publication number
- CN108074235A CN108074235A CN201711433718.XA CN201711433718A CN108074235A CN 108074235 A CN108074235 A CN 108074235A CN 201711433718 A CN201711433718 A CN 201711433718A CN 108074235 A CN108074235 A CN 108074235A
- Authority
- CN
- China
- Prior art keywords
- image
- region
- pixel
- defect
- picture
- 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
- 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
-
- 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
- G06T2207/10056—Microscopic image
- G06T2207/10061—Microscopic image from scanning electron microscope
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of carbon fiber surface defect degree methods of estimation based on region growing algorithm.First, image preprocessing is carried out to carbon fiber profile image, obtains pretreated surface defect enhancing image;Then, the ROI region of carbon fiber profile image is extracted on the basis of pretreatment image, so as to preferably find out the region of existing defects;Afterwards, the defects of being partitioned into studied carbon fiber surface picture from image using region growing algorithm region;Finally, the ratio of entire image shared by defect in image is calculated using matlab relative programs, and the number of existing defects in picture can be counted.The present invention can not only effectively be partitioned into tiny flaw existing for carbon fibre material surface, and the ratio occupied of the tiny flaw in entire image can be calculated, the quality that the carbon fibre material is assessed for professional provides a kind of reliable measurement index, plays the role of aided assessment and assesses carbon fibre material quality.
Description
Technical field
The present invention relates to the estimations of carbon fibre material surface defect, and in particular to a kind of carbon based on region growing algorithm is fine
The method of dimension table planar defect degree estimation.
Background technology
Image deflects are that occur influencing general image effect in common image or interference image information expression occur
Region or content, make researcher that can not accurately understand the information that image is conveyed, and then in the course of the research to meter
Calculation machine identifies that the image generates the part disturbed and is known as image deflects.It, can be to the surface of carbon fiber in the processing to carbon fiber
Cause defect.The reason for carbon fiber surface defect huge number, formation, is also varied.When carbon fibre composite is processed, by
In interbedded binding agent intensity and fiber intensity itself is different and fiber and the different property that are shown during cutting edge profile difference
Can, the defects of causing easily to generate tear, layering during drilling.
With the continuous development of computer application and the continuous improvement of image processing and analyzing method, with the side of machine vision
Method is split and detects to surface topography, becomes a kind of new tool of workpiece surface appearance after research mechanical processing.Vision into
The measurement of picture, integrated level is high, expansible, relatively stable, improves the flexibility of detection.Zhang Huiling, Xiao Jiayu et al. propose combination
The signal characteristic of common deficiency in carbon fiber composite material article is researched and analysed in through transmission technique C-scan and bounce technique A sweep, so as to know
Other defect.Guo Wei et al. using the modes such as Fourier transformation and wavelet transformation realize carbon fiber surface defect feature extraction and
Segmentation.Wang Keqi, Ma Xiaoming et al. carry out the segmentation of plate defect image and edge extracting using fractal theory and mathematical morphology.
Zou Lihui, white snow ice et al. are directed to using differential operator edge detection, the wealthy value segmentation of optimal iteration and morphological method has generation
Table defect is died for the sake of honour is split processing with small holes caused by worms.And such as watershed algorithm of the dividing method based on Mathematical Morphology is to faint edge
With good response, but the grey scale change that noise, body surface in image is subtle, it can all generate showing for over-segmentation
As;Partitioning algorithm based on edge detection is not suitable for multichannel image little image related to characteristic value, in image not
The image segmentation problem for having greater overlap there are the intensity value ranges of apparent each object of gray scale difference exclusive or is difficult to obtain accurate result
When noise in image signal is more, under the situations such as the gray value of target and background are very nearly the same, segment boundary information is easily lost.
On this basis, this paper presents a kind of dividing method of the carbon fiber surface defect based on region growing algorithm, there will be phase
The phenomenon that coming out with the unicom region segmentation of feature, can overcome over-segmentation simultaneously provides good boundary information and segmentation is tied
Fruit.It is chosen for seed point, the extraction for adding ROI region herein obtains the position of its seed point, is carried out further according to its position
Region growing segmentation finally acquires institute's accounting that defect area area accounts for entire figure automatically.The carbon fiber is assessed for professional
The quality of material provides a kind of reliable measurement index.
The content of the invention
The present invention provides the side that a kind of carbon fiber surface defect degree of region growing algorithm is estimated for above-mentioned background
Method, the present invention can not only effectively be partitioned into tiny flaw existing for carbon fibre material surface, and can pass through relative program
The ratio occupied of the tiny flaw in entire image is calculated, the quality that the carbon fibre material is assessed for professional provides
A kind of reliable measurement index plays the role of aided assessment and assesses carbon fibre material quality.
The technical solution adopted by the present invention is:
A kind of method of the carbon fiber surface defect degree estimation based on region growing algorithm, comprises the following steps:
A is schemed using the carbon fibre composite surface SEM under 500 times of shootings of electron microscope.Contain to what a width gave
The carbon fiber surface SEM pictures of defect area, carry out it preliminary image preprocessing, and the pretreatment of image is filtered including image
Ripple, image gray-scale transformation, image histogram equalization, image enhancement, the later pre- place of the required image enhancement of final output
Manage picture;All experiment pictures afterwards derive from image preprocessing picture herein;
B, extraction carbon fibre material surface ROI region, seed point and progress for better chosen area growth algorithm
The extraction and segmentation of the defects of accurate and effective;
C using region growing algorithm, carries out seed point selection and region growing segmentation to the picture for obtaining ROI region, carries
Take out defect area;
D, the defects of being extracted to region growing algorithm region be overlapped, utilize specific function in matlab, meter
It calculates its defect area and accounts for the area ratio of entire picture, while count by region growing algorithm extraction defect area
Number;
The method, the specific preprocessing process in the step A are as follows:
A, image filtering, homomorphic filtering basic step are as follows:
1st step, artwork can be regarded as and be made of two parts, i.e. f (x, y)=fi(x, y) fj(x, y), wherein, fiIt represents with sky
Between the different light intensity component in position, its main feature is that slowly varying, concentrate on the low frequency part of image;fjScene reflections are represented to people
The reflecting component of eye;Its feature contains the various information of scenery, and radio-frequency component enriches;
2nd step, artwork do logarithmic transformation, obtain following two additive components,
Lnf (x, y)=lnfi(x,y)+lnfj(x,y);
3rd step, logarithmic image do Fourier transformation, obtain its corresponding frequency domain representation and are:
DFT [lnf (x, y)]=DFT [lnfi(x,y)]+DFT[lnfj(x,y)]
4th step designs a frequency domain filter H (u, v), carries out the frequency domain filtering of logarithmic image;
5th step, Fourier inversion return to spatial domain logarithmic image;
6th step, fetching number obtain spatial domain filter result;
B, image gray-scale transformation, basic step are as follows:
The low gray scale value part of image is extended, shows the low more details of gray portion by the 1st step, logarithmic transformation, will
Its high gray value Partial shrinkage reduces the details of high gray scale value part, so as to achieve the purpose that emphasize the low gray portion of image;
The picture that gray scale is excessively high or gray scale is too low is modified by the 2nd step, gamma transformation, enhances contrast.Conversion is public
Formula is exactly to do product calculation to each pixel value on original image:S=cry r∈[0,1]
C, image histogram equalizes, and basic step is as follows:
1st step carries out pixel grey scale statistics, and overall pixel number N=m × n of image f is obtained
2nd step calculates each gray-scale cumulative distribution hp of image.
3rd step calculates intensity profile density, and the number of pixels of each gray scale percentage shared in whole image is obtained
Than hs (i)=hp (i)/Nf(i=0,1 ..., 255)
4th step maps gray value (equalization), g=255hs (i) i=1, and 2 ..., 255, g=0i=0
D, image enhancement, basic step are as follows:
1st step carries out two layers of wavelet decomposition
2nd step, handles decomposition coefficient, and prominent profile weakens details
3rd step, decomposition coefficient reconstruct, display enhancing image.
By above-mentioned image preprocessing, it is possible to enhanced image is obtained, thus the experiment after preferably carrying out.
The method, the step of carbon fibre material section ROI region is extracted in the step B, are as follows:
1st step reads image, and shows its attribute;
2nd step obtains binary-state threshold with threshold segmentation method, carries out binaryzation and is shown;
3rd step removes noise with operation is opened;
4th step obtains connected region, and is shown;
5th step obtains area attribute, draws center of gravity and region of interest ROI.
The method, in the step C using region growing algorithm extraction defect area the step of, are as follows:
1st step looks for starting point of the sub-pixel as growth to each region for needing to split;
2nd step, by the potting gum with sub-pixel in sub-pixel surrounding neighbors with same or similar property to kind
In region where sub-pixel;
These new pixels are continued process above, until not meeting item again by the 3rd step as new sub-pixel
The pixel of part can be included.
Algorithm realizes that process is as follows:
(1) image sequence is scanned, finds the 1st pixel belonged to not yet, if the pixel is (x0,y0);
(2) with (x0,y0) centered on, consider (x0,y0) 8 neighborhood territory pixels (x, y), if (x, y) meet growth criterion,
By (x, y) and (x0,y0) merge (in the same area), while (x, y) is pressed into storehouse;
(3) pixel is taken out from storehouse, it as (x0,y0) return to step (2);
(4) when storehouse is empty, back to step (1);
(5) step (1)-(4) are repeated when each point in image has ownership, growth terminates.
The method, in the step D, with region growing algorithm by picture it is all there are the defects of all extract
Afterwards, to it is all the defects of image carry out logic "and" operation, realize splicing/superposition of defect, an image can be obtained
In all segmentation defect area.
The method in the step D, using regionprops () function, calculates its defect area and accounts for entire picture
Area ratio;It is counted using bwlabel () function pair result images, obtains the number of defect connected region.
Compared with prior art, beneficial effects of the present invention are:
1st, the carbon fiber surface defect segmentation of method of the invention based on a kind of region growing algorithm, and then pass through related journey
Sequence calculates the ratio occupied of the tiny flaw in entire image, and the quality that the carbon fibre material is assessed for professional carries
For a kind of reliable measurement index.
2nd, by using the technology of the present invention, number of the carbon fibre material surface defect in view picture figure can be partitioned into,
By this be used as assessment carbon fibre material it is good and bad it is a kind of in a manner of;
3rd, by when number shared by the obtained carbon fiber surface defect of the present invention, auxiliary can be played from objective aspects and commented
Estimate the effect of assessed carbon fibre material quality.
Description of the drawings
Fig. 1 is frame diagram of the present invention to carbon fiber surface defect degree method of estimation.
Fig. 2 is the process and picture that the present invention carries out carbon fiber sectional view image preprocessing.
Fig. 3 is to carry out the process of ROI region extraction and result picture to pretreatment picture;(A) extraction process;(B) it is final
As a result.
Fig. 4 is that the present invention carries out carbon fiber sectional view region growth and the process and picture that are overlapped.
Fig. 5 is that the present invention calculates defect connected region area and calculates the result of defect institute accounting;(A) defect area is superimposed
The gross area;(B) defect area accounting.
Fig. 6 is the result that the present invention calculates defect connected region number;(A) image is enhanced;(B) defect area is superimposed;(C)
Defect area number.
Table 1 is that Electronic Speculum shooting exemplar sampling is taken the surface sample in 500 times of five faces of Electronic Speculum to be tested, drawn respectively
Layering and the average value of tearing defect institute accounting.
Specific embodiment
Below in conjunction with specific embodiment, the present invention is described in detail.
With reference to Fig. 1,3, the realization flow of the method for the present invention is as follows:
A kind of method of the carbon fiber surface defect degree estimation based on region growing algorithm, comprises the following steps:
A is schemed using the carbon fibre composite surface SEM under 500 times of shootings of electron microscope.Contain to what a width gave
The carbon fiber surface SEM pictures of defect area, carry out it preliminary image preprocessing, and the pretreatment of image is filtered including image
Ripple, image gray-scale transformation, image histogram equalization, image enhancement, the later pre- place of the required image enhancement of final output
Manage picture;All experiment pictures afterwards derive from image preprocessing picture herein;
B, extraction carbon fibre material surface ROI region, seed point and progress for better chosen area growth algorithm
The extraction and segmentation of the defects of accurate and effective;
C using region growing algorithm, carries out seed point selection and region growing segmentation to the picture for obtaining ROI region, carries
Take out defect area;
D, the defects of being extracted to region growing algorithm region be overlapped, utilize specific function in matlab, meter
It calculates its defect area and accounts for the area ratio of entire picture, while count by region growing algorithm extraction defect area
Number;
The method, the specific preprocessing process in the step A are as follows:
A, image filtering, homomorphic filtering basic step are as follows:
1st step, artwork can be regarded as and be made of two parts, i.e. f (x, y)=fi(x, y) fj(x, y), wherein, fiIt represents with sky
Between the different light intensity component in position, its main feature is that slowly varying, concentrate on the low frequency part of image;fjScene reflections are represented to people
The reflecting component of eye;Its feature contains the various information of scenery, and radio-frequency component enriches;
2nd step, artwork do logarithmic transformation, obtain following two additive components,
Lnf (x, y)=lnfi(x,y)+lnfj(x,y);
3rd step, logarithmic image do Fourier transformation, obtain its corresponding frequency domain representation and are:
DFT [lnf (x, y)]=DFT [lnfi(x,y)]+DFT[lnfj(x,y)]
4th step designs a frequency domain filter H (u, v), carries out the frequency domain filtering of logarithmic image;
5th step, Fourier inversion return to spatial domain logarithmic image;
6th step, fetching number obtain spatial domain filter result;
B, image gray-scale transformation, basic step are as follows:
The low gray scale value part of image is extended, shows the low more details of gray portion by the 1st step, logarithmic transformation, will
Its high gray value Partial shrinkage reduces the details of high gray scale value part, so as to achieve the purpose that emphasize the low gray portion of image;
The picture that gray scale is excessively high or gray scale is too low is modified by the 2nd step, gamma transformation, enhances contrast.Conversion is public
Formula is exactly to do product calculation to each pixel value on original image:S=cry r∈[0,1]
C, image histogram equalizes, and basic step is as follows:
1st step carries out pixel grey scale statistics, and overall pixel number N=m × n of image f is obtained
2nd step calculates each gray-scale cumulative distribution hp of image.
3rd step calculates intensity profile density, and the number of pixels of each gray scale percentage shared in whole image is obtained
Than hs (i)=hp (i)/Nf(i=0,1 ..., 255)
4th step maps gray value (equalization), g=255hs (i) i=1, and 2 ..., 255, g=0i=0
D, image enhancement, basic step are as follows:
1st step carries out two layers of wavelet decomposition
2nd step, handles decomposition coefficient, and prominent profile weakens details
3rd step, decomposition coefficient reconstruct, display enhancing image.
By above-mentioned image preprocessing, it is possible to enhanced image is obtained, thus the experiment after preferably carrying out.
The method, the step of carbon fibre material section ROI region is extracted in the step B, are as follows:
1st step reads image, and shows its attribute;
2nd step obtains binary-state threshold with threshold segmentation method, carries out binaryzation and is shown;
3rd step removes noise with operation is opened;
4th step obtains connected region, and is shown;
5th step obtains area attribute, draws center of gravity and region of interest ROI.
The method, in the step C using region growing algorithm extraction defect area the step of, are as follows:
1st step looks for starting point of the sub-pixel as growth to each region for needing to split;
2nd step, by the potting gum with sub-pixel in sub-pixel surrounding neighbors with same or similar property to kind
In region where sub-pixel;
These new pixels are continued process above, until not meeting item again by the 3rd step as new sub-pixel
The pixel of part can be included.
Algorithm realizes that process is as follows:
(1) image sequence is scanned, finds the 1st pixel belonged to not yet, if the pixel is (x0,y0);
(2) with (x0,y0) centered on, consider (x0,y0) 8 neighborhood territory pixels (x, y), if (x, y) meet growth criterion,
By (x, y) and (x0,y0) merge (in the same area), while (x, y) is pressed into storehouse;
(3) pixel is taken out from storehouse, it as (x0,y0) return to step (2);
(4) when storehouse is empty, back to step (1);
(5) step (1)-(4) are repeated when each point in image has ownership, growth terminates.
The method, in the step D, with region growing algorithm by picture it is all there are the defects of all extract
Afterwards, to it is all the defects of image carry out logic "and" operation, realize splicing/superposition of defect, an image can be obtained
In all segmentation defect area.
The method in the step D, using regionprops () function, calculates its defect area and accounts for entire picture
Area ratio;It is counted using bwlabel () function pair result images, obtains the number of defect connected region.
The realization flow of the method for the present invention is as follows:
(1) image preprocessing is carried out to image to be assessed
(2) image is filtered with Butterworth homomorphic filtering
(3) image gray-scale transformation is carried out to image with logarithmic transformation and gamma transformation
(4) histogram equalization is carried out to the image after greyscale transformation
(5) reconstructed by decomposition coefficient, obtain enhancing image
(6) (ROI) extracted region interested is carried out to pretreatment image
(7) after pre-processing, carbon fiber surface defect is partitioned into region growing algorithm
(8) to being partitioned into the defects of, is overlapped
(9) the when number of defect is acquired shared by defect.
Fig. 2 present invention pre-processes carbon fibre material section, including image filtering, image gray-scale transformation, image
The methods of histogram equalization, image enhancement, eliminates the noise in image, the defects of follow-up to be facilitated to extract.
The result figure that Fig. 3 illustrates the process of present invention extraction ROI region and obtains.Including with threshold segmentation method
Obtain binary-state threshold, using opening operation removal noise and obtaining connected region, the final area attribute that obtains simultaneously draws center of gravity
And region of interest ROI.
Fig. 4 illustrates to carry out carbon fiber sectional view region growth and the process and image that are overlapped, by original
The processing such as the progressively filtering of input picture, enhancing, then the progressively processing to treated image carries out region growing algorithm, afterwards
The superposition of image is carried out, can thus clearly show that out defect present in a width carbon fiber profile image, and then after being
Continuous research provides foundation.
Fig. 5 shows that computer program calculates defect connected region area, and then calculates the result of defect proportion.It is right
In the carbon fiber profile image of a unknown existing defects, by a series of processing before, using region growing algorithm
To it is existing the defects of extract and split, meanwhile, computer program can successfully show the area of existing defect, into
And the area that defect area area accounts for whole image can be calculated, i.e. defect proportion.
Fig. 6 is the result that the present invention calculates defect connected region number.Defect area is carried out by region growing algorithm
After segmentation extraction, region is overlapped connection the defects of to all extractions, you can shows in unknown images and deposits in piece image
All defect region, and then can smoothly calculate with computer program the number of connected region in handled image.
Table 1 is that Electronic Speculum shooting exemplar sampling is taken the surface sample in 500 times of five faces of Electronic Speculum to be tested, drawn respectively
Layering and the average value of tearing defect institute accounting.Can clearly showing that out defect in piece image by average value, (tear divides
Layer) shared by ratio, and then can accurately weigh the quality of studied carbon fibre material quality.
Table 1
Surface sample | Tear % | It is layered % |
500 first faces of Electronic Speculum | 0.13 | 0.07 |
500 second faces of Electronic Speculum | 0.41 | 0.11 |
The 3rd face of Electronic Speculum 500 | 0.16 | 0.06 |
The 4th face of Electronic Speculum 500 | 0.09 | 0.12 |
The 5th face of Electronic Speculum 500 | 0.23 | 0.35 |
The foregoing is only a preferred embodiment of the present invention, protection scope of the present invention is without being limited thereto, it is any ripe
Those skilled in the art are known in the technical scope of present disclosure, the letter for the technical solution that can be become apparent to
Altered or equivalence replacement are each fallen in protection scope of the present invention.
Claims (6)
- A kind of 1. method of the carbon fiber surface defect degree estimation based on region growing algorithm, which is characterized in that including following Step:A is schemed using the carbon fibre composite surface SEM under 500 times of shootings of electron microscope.To a width give containing defective The carbon fiber surface SEM pictures in region, carry out it preliminary image preprocessing, and the pretreatment of image includes image filtering, figure As greyscale transformation, image histogram equalization, image enhancement, the later pretreatment figure of the required image enhancement of final output Piece;All experiment pictures afterwards derive from image preprocessing picture herein;B, extraction carbon fibre material surface ROI region, for the seed point of better chosen area growth algorithm and progress is accurate The extraction and segmentation of effectively the defects of;C using region growing algorithm, carries out seed point selection and region growing segmentation to the picture for obtaining ROI region, extracts Defect area;D, the defects of being extracted to region growing algorithm region be overlapped, using specific function in matlab, calculate it Defect area accounts for the area ratio of entire picture, while counts the number by region growing algorithm extraction defect area.
- 2. according to the method described in claim 1, it is characterized in that, the specific preprocessing process in the step A is as follows:A, image filtering, homomorphic filtering basic step are as follows:1st step, artwork can be regarded as and be made of two parts, i.e. f (x, y)=fi(x, y) fj(x, y), wherein, fiIt represents with space bit Different light intensity components is put, its main feature is that it is slowly varying, concentrate on the low frequency part of image;fjScene reflections are represented to human eye Reflecting component;Its feature contains the various information of scenery, and radio-frequency component enriches;2nd step, artwork do logarithmic transformation, obtain following two additive components,Lnf (x, y)=lnfi(x,y)+lnfj(x,y);3rd step, logarithmic image do Fourier transformation, obtain its corresponding frequency domain representation and are:DFT [lnf (x, y)]=DFT [lnfi(x,y)]+DFT[lnfj(x,y)]4th step designs a frequency domain filter H (u, v), carries out the frequency domain filtering of logarithmic image;5th step, Fourier inversion return to spatial domain logarithmic image;6th step, fetching number obtain spatial domain filter result;B, image gray-scale transformation, basic step are as follows:The low gray scale value part of image is extended, shows the low more details of gray portion by the 1st step, logarithmic transformation, it is high Gray value Partial shrinkage reduces the details of high gray scale value part, so as to achieve the purpose that emphasize the low gray portion of image;The picture that gray scale is excessively high or gray scale is too low is modified by the 2nd step, gamma transformation, enhances contrast.Transformation for mula is just It is that product calculation is done to each pixel value on original image:S=cry r∈[0,1]C, image histogram equalizes, and basic step is as follows:1st step carries out pixel grey scale statistics, and overall pixel number N=m × n of image f is obtained2nd step calculates each gray-scale cumulative distribution hp of image.3rd step calculates intensity profile density, and the number of pixels of each gray scale percentage shared in whole image, hs is obtained (i)=hp (i)/Nf(i=0,1 ..., 255)4th step maps gray value (equalization), g=255hs (i) i=1, and 2 ..., 255, g=0i=0D, image enhancement, basic step are as follows:1st step carries out two layers of wavelet decomposition2nd step, handles decomposition coefficient, and prominent profile weakens details3rd step, decomposition coefficient reconstruct, display enhancing image.By above-mentioned image preprocessing, it is possible to enhanced image is obtained, thus the experiment after preferably carrying out.
- 3. according to the method described in claim 1, it is characterized in that, carbon fibre material section ROI region is extracted in the step B The step of it is as follows:1st step reads image, and shows its attribute;2nd step obtains binary-state threshold with threshold segmentation method, carries out binaryzation and is shown;3rd step removes noise with operation is opened;4th step obtains connected region, and is shown;5th step obtains area attribute, draws center of gravity and region of interest ROI.
- 4. method according to claim 1, which is characterized in that defect area is extracted using region growing algorithm in the step C The step of domain, is as follows:1st step looks for starting point of the sub-pixel as growth to each region for needing to split;2nd step, by the potting gum with sub-pixel in sub-pixel surrounding neighbors with same or similar property to seed picture In region where plain;These new pixels are continued process above, until not meeting condition again by the 3rd step as new sub-pixel Pixel can be included.Algorithm realizes that process is as follows:(1) image sequence is scanned, finds the 1st pixel belonged to not yet, if the pixel is (x0,y0);(2) with (x0,y0) centered on, consider (x0,y0) 8 neighborhood territory pixels (x, y), if (x, y) meet growth criterion, will (x, Y) with (x0,y0) merge (in the same area), while (x, y) is pressed into storehouse;(3) pixel is taken out from storehouse, it as (x0,y0) return to step (2);(4) when storehouse is empty, back to step (1);(5) step (1)-(4) are repeated when each point in image has ownership, growth terminates.
- 5. method according to claim 1, which is characterized in that in the step D, will own with region growing algorithm in picture There are the defects of all extract after, to it is all the defects of image carry out logic "and" operation, realize the splicing of defect/folded Add, segmentation defect area all in an image can be obtained.
- 6. method according to claim 1, which is characterized in that in the step D, using regionprops () function, calculate Its defect area accounts for the area ratio of entire picture;It is counted using bwlabel () function pair result images, obtains defect The number of connected region.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711433718.XA CN108074235A (en) | 2017-12-26 | 2017-12-26 | Carbon fiber surface defect degree method of estimation based on region growing algorithm |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711433718.XA CN108074235A (en) | 2017-12-26 | 2017-12-26 | Carbon fiber surface defect degree method of estimation based on region growing algorithm |
Publications (1)
Publication Number | Publication Date |
---|---|
CN108074235A true CN108074235A (en) | 2018-05-25 |
Family
ID=62155236
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201711433718.XA Pending CN108074235A (en) | 2017-12-26 | 2017-12-26 | Carbon fiber surface defect degree method of estimation based on region growing algorithm |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108074235A (en) |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108765467A (en) * | 2018-06-21 | 2018-11-06 | 西安胡门网络技术有限公司 | A kind of Moving small targets detection tracking based on video detection |
CN109870460A (en) * | 2019-03-24 | 2019-06-11 | 哈尔滨理工大学 | A kind of composite material battery case surfaces quality determining method based on machine vision |
CN109959664A (en) * | 2019-04-10 | 2019-07-02 | 长春禹衡光学有限公司 | Contamination detection method, device and the readable storage medium storing program for executing of absolute grating ruler |
CN110288540A (en) * | 2019-06-04 | 2019-09-27 | 东南大学 | A kind of online imaging standards method of carbon-fibre wire radioscopic image |
CN110400319A (en) * | 2019-07-16 | 2019-11-01 | 东华大学 | A kind of spinning cake greasy dirt partitioning algorithm based on domain division method |
CN111489351A (en) * | 2020-04-20 | 2020-08-04 | 东南大学 | Video analysis method for measuring surface segregation degree based on stack image |
CN112801938A (en) * | 2020-12-30 | 2021-05-14 | 南通景康橡塑有限公司 | Method and device for intelligently detecting quality of rubber and plastic material |
CN112954304A (en) * | 2021-01-18 | 2021-06-11 | 湖北经济学院 | Mura defect evaluation method and system for display panel and readable storage medium |
CN117314899A (en) * | 2023-11-28 | 2023-12-29 | 深圳市烯碳复合材料有限公司 | Carbon fiber plate quality detection method based on image characteristics |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1912927A (en) * | 2006-08-25 | 2007-02-14 | 西安理工大学 | Semi-automatic partition method of lung CT image focus |
-
2017
- 2017-12-26 CN CN201711433718.XA patent/CN108074235A/en active Pending
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1912927A (en) * | 2006-08-25 | 2007-02-14 | 西安理工大学 | Semi-automatic partition method of lung CT image focus |
Non-Patent Citations (8)
Title |
---|
SAHOO A K ET AL.: "A new approach for parallel region growing algorithm in image segmentation using MATLAB on GPU architecture", 《IEEE INTERNATIONAL CONFERENCE ON COMPUTER GRAPHICS,VISION AND INFORMATION SECURITY 2015》 * |
冯琪智 等: "基于热图重构区域生长算法的碳纤维增强复合材料脱粘缺陷检测", 《无损检测》 * |
吴海滨 等: "基于二维OTSU选取种子点的区域生长图像分割", 《大气与环境光学学报》 * |
娄春华: "《聚合物结构与性能》", 31 May 2016 * |
朱良漪 等: "《分析仪器手册》", 31 May 1997 * |
王明强 等: "《现代机械设计理论与应用》", 31 December 2011 * |
陆系群 等: "《图像处理原理、技术与算法》", 31 August 2001 * |
陈志新: "《对偶树复小波分析及其应用》", 30 April 2014 * |
Cited By (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108765467A (en) * | 2018-06-21 | 2018-11-06 | 西安胡门网络技术有限公司 | A kind of Moving small targets detection tracking based on video detection |
CN109870460A (en) * | 2019-03-24 | 2019-06-11 | 哈尔滨理工大学 | A kind of composite material battery case surfaces quality determining method based on machine vision |
CN109959664A (en) * | 2019-04-10 | 2019-07-02 | 长春禹衡光学有限公司 | Contamination detection method, device and the readable storage medium storing program for executing of absolute grating ruler |
CN110288540B (en) * | 2019-06-04 | 2021-07-06 | 东南大学 | Carbon fiber wire X-ray image online imaging standardization method |
CN110288540A (en) * | 2019-06-04 | 2019-09-27 | 东南大学 | A kind of online imaging standards method of carbon-fibre wire radioscopic image |
CN110400319A (en) * | 2019-07-16 | 2019-11-01 | 东华大学 | A kind of spinning cake greasy dirt partitioning algorithm based on domain division method |
CN111489351A (en) * | 2020-04-20 | 2020-08-04 | 东南大学 | Video analysis method for measuring surface segregation degree based on stack image |
CN111489351B (en) * | 2020-04-20 | 2022-11-18 | 东南大学 | Video analysis method for measuring surface segregation degree based on stack image |
CN112801938A (en) * | 2020-12-30 | 2021-05-14 | 南通景康橡塑有限公司 | Method and device for intelligently detecting quality of rubber and plastic material |
CN112801938B (en) * | 2020-12-30 | 2021-12-14 | 南通景康橡塑有限公司 | Method and device for intelligently detecting quality of rubber and plastic material |
CN112954304A (en) * | 2021-01-18 | 2021-06-11 | 湖北经济学院 | Mura defect evaluation method and system for display panel and readable storage medium |
CN117314899A (en) * | 2023-11-28 | 2023-12-29 | 深圳市烯碳复合材料有限公司 | Carbon fiber plate quality detection method based on image characteristics |
CN117314899B (en) * | 2023-11-28 | 2024-03-08 | 深圳市烯碳复合材料有限公司 | Carbon fiber plate quality detection method based on image characteristics |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108074235A (en) | Carbon fiber surface defect degree method of estimation based on region growing algorithm | |
CN107194919B (en) | Mobile phone screen defect detection method based on regular texture background reconstruction | |
CN108053417B (en) | lung segmentation device of 3D U-Net network based on mixed rough segmentation characteristics | |
CN104318546B (en) | Multi-scale analysis-based greenhouse field plant leaf margin extraction method and system | |
Kanwal et al. | Region based adaptive contrast enhancement of medical X-ray images | |
CN106327451A (en) | Image restorative method of ancient animal fossils | |
CN109544571A (en) | A kind of metallic phase image edge detection method based on mathematical morphology | |
CN101976440B (en) | Sobel operator-based extraction method of profile and detail composite characteristic vector used for representing fabric texture | |
CN108492268A (en) | Enhancement algorithm for low-illumination image based on wavelet coefficient fusion | |
Hazarika et al. | A new breast border extraction and contrast enhancement technique with digital mammogram images for improved detection of breast cancer | |
Vikhe et al. | Contrast enhancement in mammograms using homomorphic filter technique | |
CN106780718A (en) | A kind of three-dimensional rebuilding method of paleontological fossil | |
Tripathy et al. | Performance observation of mammograms using an improved dynamic window based adaptive median filter | |
CN115100077A (en) | Novel image enhancement method and device | |
CN107392204A (en) | A kind of galactophore image microcalcifications automatic checkout system and method | |
CN101976441B (en) | Method for extracting Sobel operator filtering profile for representing fabric texture and fractal detail mixed characteristic vector | |
Sofou et al. | Generalized flooding and multicue PDE-based image segmentation | |
Biswas et al. | A model of noise reduction using Gabor Kuwahara Filter | |
Song et al. | An adaptive real-time video defogging method based on context-sensitiveness | |
CN109377461A (en) | A kind of breast X-ray image self-adapting enhancement method based on NSCT | |
Budhiraja et al. | Effect of pre-processing on MST based infrared and visible image fusion | |
Qu et al. | LEUGAN: low-light image enhancement by unsupervised generative attentional networks | |
Hu et al. | Infrared and visible image fusion based on multiscale decomposition with Gaussian and co-occurrence filters | |
Vikhe et al. | A wavelet and adaptive threshold-based contrast enhancement of masses in mammograms for visual screening | |
Elhefnawy et al. | Effective visibility restoration and enhancement of air polluted images with high information fidelity |
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: 20180525 |
|
RJ01 | Rejection of invention patent application after publication |