CN107977954A - Textile flaw detection method based on local optimum analysis - Google Patents
Textile flaw detection method based on local optimum analysis Download PDFInfo
- Publication number
- CN107977954A CN107977954A CN201710946334.1A CN201710946334A CN107977954A CN 107977954 A CN107977954 A CN 107977954A CN 201710946334 A CN201710946334 A CN 201710946334A CN 107977954 A CN107977954 A CN 107977954A
- Authority
- CN
- China
- Prior art keywords
- textile
- block
- image block
- image
- template
- 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.)
- Withdrawn
Links
- 239000004753 textile Substances 0.000 title claims abstract description 70
- 238000001514 detection method Methods 0.000 title claims abstract description 30
- 238000004458 analytical method Methods 0.000 title claims abstract description 15
- 238000000034 method Methods 0.000 claims abstract description 30
- 238000004364 calculation method Methods 0.000 claims description 4
- 238000000205 computational method Methods 0.000 claims description 3
- 238000002372 labelling Methods 0.000 claims description 3
- 238000005192 partition Methods 0.000 claims description 3
- 238000009941 weaving Methods 0.000 claims description 2
- 238000009792 diffusion process Methods 0.000 claims 1
- 230000000737 periodic effect Effects 0.000 abstract description 4
- 238000007689 inspection Methods 0.000 abstract description 3
- 230000007547 defect Effects 0.000 description 7
- 238000005516 engineering process Methods 0.000 description 4
- 238000012360 testing method Methods 0.000 description 3
- 230000002950 deficient Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000012372 quality testing Methods 0.000 description 2
- 239000007787 solid Substances 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000001186 cumulative effect Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 239000004744 fabric Substances 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 230000003595 spectral effect Effects 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
- 238000009987 spinning Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
- G06T7/001—Industrial image inspection using an image reference approach
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20021—Dividing image into blocks, subimages or windows
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
- G06T2207/30124—Fabrics; Textile; Paper
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30168—Image quality inspection
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Quality & Reliability (AREA)
- Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
- Treatment Of Fiber Materials (AREA)
Abstract
The present invention relates to textile inspection technical field, is based especially on the textile flaw detection method that local optimum is analyzed.The detection method comprises the following steps：1) textile images to be detected containing mechanical periodicity pattern are inputted；2) the cycle template size of pattern is determined；3) piecemeal is carried out to image according to template size；4) Markov random field models are established to the image block of piecemeal；5) local energy of each image block is calculated, changes the mark of block so that the energy reaches stable, and then diffuses to global optimum by local optimum；6) positioning of flaw is completed according to final Label Field.The present invention provides a kind of textile flaw detection method based on local optimum analysis, and image is cut the size by determining periodic pattern and piecemeal, reduces computation complexity, improves detection rates, while this method has universality to flaw type.
Description
Technical field
The present invention relates to textile inspection technical field, is based especially on the textile Defect Detection side that local optimum is analyzed
Method.
Background technology
The economic benefit of textile is determined that textile of fine qualities can bring income, and contain flaw by its quality
Defect ware can then bring economic loss, traditional manual detection mode is the scoring of the experience, textile according to testing staff
With the standard such as comment to be evaluated to the quality of textile.This mode detection speed is low, and omission factor is higher, it is therefore desirable to sends out
Open up quick, accurate and unsupervised textile flaw detection method.
The fabric type that textile Defect Detection is directed to now, can be divided into two classes：The first kind is simple in structure, is not contained
Complicated pattern, is mostly the textile of pure color；Second class is then to have complex pattern-information, and pattern has periodically.
Quality testing for first kind pure color textile images has had many ripe technologies and algorithm, and comparing has
It is representational to have statistic law, Spectrum Method, coaching method, Structure Method and modelling etc..Wherein, statistic law and spectral method can not be directed to the flaw
The larger textile images of defect area are detected；Coaching method needs training parameter in large quantities, and the time consumed and cost are all
It is very high；Structure Method is very high to the requirement of sample image texture, does not have universality to flaw type.In these techniques, modelling energy
It is enough that Defect Detection problem is converted into assumed statistical inspection problem, using limited parameter describe a pixel in image with
The statistic correlation of pixel in its adjacent area, can effectively describe the architectural characteristic and statistical property of texture.
Contain the quality testing of the complicated textile images of cycle primitive pattern for the second class, ripe technology is relatively
It is few, and existing technology is required to the parameter of training sample, required time and cost are high, and testing result is also general.
The content of the invention
The technical problem to be solved in the present invention is：In order to solve, the existing detection method cycle is long, of high cost and effect is poor
Deficiency, the present invention provides it is a kind of based on local optimum analysis textile flaw detection method, pass through determine periodic pattern
Size image is cut and piecemeal, reduce computation complexity, improve detection rates, while this method is to flaw class
Type has universality.
The technical solution adopted by the present invention to solve the technical problems is：
A kind of textile flaw detection method based on local optimum analysis, comprises the following steps：
1) textile images to be detected containing mechanical periodicity pattern are inputted；
2) the cycle template size of pattern is determined；
3) piecemeal is carried out to image according to template size；
4) Markov random field models are established to the image block of piecemeal；
5) local energy of each image block is calculated, changes the mark of block so that the energy reaches stable, Jin Eryou
Local optimum diffuses to global optimum；
6) positioning of flaw is completed according to final Label Field.
Specifically, the method for the cycle template size of the definite pattern is to take step in the horizontal direction of textile images
Long c and take step-length r to carry out even partition to textile images in the vertical direction of textile images, calculate adjacent difference between two pieces
Summation that is different and calculating all differences, and ask for according to the minimum of difference summation the template size of textile images.
Specifically, the computational methods of the template size for asking for textile images are to make step-length c be equal to textile images
Horizontal direction image length carry out the solution of r, then solve c, or the height pattern image width for making step-length r be equal to textile images
Degree carries out the solution of c, then solves r, and obtains the optimum value of r and c as template size.
Specifically, it is described to be to the method for image progress piecemeal according to template size, according to template size and textile figure
The length and width of picture cut textile images；The textile after cutting is averagely divided into according to template size some
The identical image block of a size.
Specifically, the method that the image block to piecemeal establishes Markov random field models is, each image block be with
The minimum calculation unit on airport；Current block forms a neighborhood system with its four neighborhood images block；The property of current block can be by
The property representation of four neighborhoods；A mark is assigned to each image block, initial markers are 0, using the markd set of institute as
Label Field；All image blocks are overlapped, generate three-dimensional model, calculate the median of the model, obtained result is to treat
Survey textile images global template, the template be defaulted as it is flawless, its be labeled as 0.
Specifically, the method for the local energy for calculating each image block is, identical with overall situation template for mark
Image block, counts in its four neighborhood and marks identical image block with it, it is identical with its four neighborhoods internal labeling to calculate the image block
The average of similarity between block；For marking the image block different from global template, its local energy is equal to mark and global mould
The Energy maximum value of the identical image block of plate.
Specifically, the method for the mark for changing block is the similar system calculated between all image blocks and global template
Number；A decision value λ is taken to be traveled through between (0,1)；Change image block of the similarity factor less than λ to mark；If local energy after changing mark
Amount reduces, and illustrates to mark iteration accurate, otherwise mistake, changes back former mark.
Specifically, the method that the final Label Field of the basis completes the positioning of flaw is, labeled as 0 image block
Flawless block is regarded as, flaw block has been regarded as labeled as 1 image block.
The beneficial effects of the invention are as follows：The present invention provides a kind of textile Defect Detection side based on local optimum analysis
Method, image is cut the size by determining periodic pattern and piecemeal, reduces computation complexity, improves detection speed
Rate, while this method has universality to flaw type.
Brief description of the drawings
The present invention is further described with reference to the accompanying drawings and examples.
Fig. 1 is the detection method FB(flow block) of the present invention；
Fig. 2 is total difference curve figure of the vertical direction of the present invention；
Fig. 3 is total difference curve figure of the horizontal direction of the present invention；
Fig. 4 is the Markov random field models neighborhood systems of the present invention；
Fig. 5 is the calculating image block of the present invention and global template similarity factor schematic diagram；
Embodiment
In conjunction with the accompanying drawings, the present invention is further explained in detail.
Fig. 1 be the present invention detection method FB(flow block), Fig. 2 be the present invention vertical direction total difference curve figure, Fig. 3
It is total difference curve figure of the horizontal direction of the present invention, Fig. 4 is the Markov random field models neighborhood systems of the present invention, and Fig. 5 is
The calculating image block of the present invention and global template similarity factor schematic diagram.
A kind of textile flaw detection method based on local optimum analysis, it is characterized in that, comprise the following steps：
1) textile images to be detected containing mechanical periodicity pattern are inputted；
2) the cycle template size of pattern is determined；
3) piecemeal is carried out to image according to template size；
4) Markov random field models are established to the image block of piecemeal；
5) local energy of each image block is calculated, changes the mark of block so that the energy reaches stable, Jin Eryou
Local optimum diffuses to global optimum；
6) positioning of flaw is completed according to final Label Field.
The method of the cycle template size of the definite pattern is, the horizontal direction of textile images take step-length c and
The vertical direction of textile images takes step-length r to carry out even partition to textile images, calculates difference and meter between adjacent two pieces
The summation of all differences is calculated, and the template size of textile images is asked for according to the minimum of difference summation.It is described to ask for weaving
The computational methods of the template size of product image are that the horizontal direction image length for making step-length c be equal to textile images carries out asking for r
Solution, then c is solved, or the vertical direction picture traverse for making step-length r be equal to textile images carries out the solution of c, then r is solved, and obtain
The optimum value of r and c is taken as template size.It is described to be to the method for image progress piecemeal according to template size, it is big according to template
Small and textile images length and width cut textile images；The textile after cutting is put down according to template size
It is divided into the identical image block of several sizes.The method that the image block to piecemeal establishes Markov random field models
It is that each image block is the minimum calculation unit of random field；Current block forms a neighborhood system with its four neighborhood images block；When
Preceding piece of property can be by the property representation of four neighborhoods；A mark is assigned to each image block, initial markers are 0, by institute
Markd set is used as Label Field；All image blocks are overlapped, generate three-dimensional model, calculate the median of the model,
Obtained result is the global template of textile images to be measured, which is defaulted as flawless, it is labeled as 0.It is described to calculate often
The method of the local energy of one image block is, for marking the image block identical with global template, count its four neighborhood it is interior and its
Identical image block is marked, calculates the average of similarity between the image block block identical with its four neighborhoods internal labeling；For mark
Remember the image block different from global template, the energy that its local energy is equal to the mark image block identical with global template is maximum
Value.The method of the mark for changing block is the similarity factor calculated between all image blocks and global template；Take a decision value
λ is traveled through between (0,1)；Change image block of the similarity factor less than λ to mark；If changing local energy after mark reduces, illustrate mark
Remember that iteration is accurate, on the contrary mistake, change back former mark.
The method that the final Label Field of the basis completes the positioning of flaw is, labeled as 0 image block regard as it is flawless
Block, flaw block has been regarded as labeled as 1 image block.
Determine the size of textile images piecemeal template to be measured, contain basic pattern for one, size is the spinning of M × N
Fabric image I, will determine the basic pattern template that size is m × n, Schilling template width m=M, form height n is in [1, N/2]
Interior takes a numerical value d, forms one and treats solid plate, will should treat that solid plate is divided into by IBlock, is respectivelyCalculate adjacent two pieces of difference and the total difference f of cumulative read group total；By the way that d is traveled through in [1, N/2], look for
Go out so that total difference f reaches extreme value, and the d during area minimum of piecemeal is the height n of image template, template width m is really
Process is determined, so that it is determined that size is the base pattern template of m × n.Textile images I is cut simultaneously by the template
It is divided into block, note image block collection is combined into B={ b1,b2,…bp×q}。
Wherein total difference f calculation formula are as follows：
Since image has periodically, so when d is the integral multiple of cycle template, total difference f will obtain minimum,
So the template of area minimum is only optimal template in all length and width for making f obtain minimum.
As shown in Figure 2, for complicated textile images in the horizontal direction and vertical direction total difference curve figure.By attached drawing
Known to 2, the point that form height obtains minimum is respectively 21,42,63,84,106, and template width obtains the point difference of minimum
For 16,33,49,66,83,100.Since pattern has periodically, so template should select area minimum, so such pattern
This template size is about 21 × 16.
For according to template by the result after textile images piecemeal.For containing image defective, also according to template into
Row piecemeal, since flaw area is less than the 50% of textile images area, so all image blocks are overlapped, forms one
Three-dimensional model, the median for calculating the three-dimensional model can obtain flawless global template, which is denoted as 0.Calculate institute
There is the similarity degree between image block and global template, form similarity factor matrix.
As shown in Figure 3, central block s and its four neighborhoods block r constitutes the neighborhood system of s.It is right in Markov random fields
It is as follows in the mark image block identical with global template, its energy balane formula：
Wherein, xi and xr is current center block and the property of its neighborhood block, and yi and the corresponding marks of yr, V1 and V2 are respectively
The single order and second order potential function of central block, i (k, l), GT (k, l), r (k, l) are respectively that central block, global template and neighborhood block exist
The pixel value at position (k, l) place.H is the number identical with current block mark in four neighborhoods, h ∈ { 1,2,3,4 }.
And for being replaced labeled as the image block for having the flaw, its energy by the maximum labeled as flawless image block energy,
I.e.
Us(ys≠yGT)=maxUs(ys=yGT) (5)
A decision value λ is made to be traveled through in the range of (0,1) with certain step-length, when a certain image block and global template
Similarity factor when being less than λ, change the mark of the block, the local energy after changing mark calculated, if changing the image block
The energy summation of entire image is less than the energy summation before changing after mark, just illustrates that image block mark iteration is accurate, on the contrary
Then mark constant.Solution MRF model parameter problems can be converted into using similarity relation change image tagged and seek all images
What block was formed under a certain decision value causes local energy summation to reach minimum Label Field, and formula is such as shown in (6)：
The region of flaw is determined by final Label Field：Flawless block is considered labeled as 0 image block, labeled as 1
Image block is considered to contain block defective, in testing result, only can show flaw block, final to solve to contain periodic pattern
The Defect Detection problem of textile images.
It should be appreciated that although the present specification is described in terms of embodiments, not each embodiment only includes one
A independent technical solution, this narrating mode of specification is only that those skilled in the art will should say for clarity
For bright book as an entirety, the technical solution in each embodiment may also be suitably combined to form those skilled in the art can
With the other embodiment of understanding.
Claims (8)
1. a kind of textile flaw detection method based on local optimum analysis, it is characterized in that, comprise the following steps：
1) textile images to be detected containing mechanical periodicity pattern are inputted；
2) the cycle template size of pattern is determined；
3) piecemeal is carried out to image according to template size；
4) Markov random field models are established to the image block of piecemeal；
5) local energy of each image block is calculated, changes the mark of block so that the energy reaches stable, and then by part
Optimal diffusion is to global optimum；
6) positioning of flaw is completed according to final Label Field.
2. the textile flaw detection method according to claim 1 based on local optimum analysis, it is characterised in that：It is described
Determining the method for the cycle template size of pattern is, step-length c is taken and in textile images in the horizontal direction of textile images
Vertical direction takes step-length r to carry out even partition to textile images, calculates the difference between adjacent two pieces and calculates all differences
Summation, and ask for according to the minimum of difference summation the template size of textile images.
3. the textile flaw detection method according to claim 2 based on local optimum analysis, it is characterised in that：It is described
Asking for the computational methods of the template size of textile images is, makes step-length c be equal to the horizontal direction image length of textile images
The solution of r is carried out, then solves c, or the vertical direction picture traverse for making step-length r be equal to textile images carries out the solution of c, then ask
R is solved, and obtains the optimum value of r and c as template size.
4. the textile flaw detection method according to claim 1 based on local optimum analysis, it is characterised in that：It is described
Carrying out the method for piecemeal to image according to template size is, according to the length and width of template size and textile images to weaving
Product image is cut；Textile after cutting is averagely divided into by the identical image block of several sizes according to template size.
5. the textile flaw detection method according to claim 1 based on local optimum analysis, it is characterised in that：It is described
The method of Markov random field models is established to the image block of piecemeal is, each image block is the minimum calculation unit of random field；
Current block forms a neighborhood system with its four neighborhood images block；The property of current block can be by the property representation of four neighborhoods；It is right
Each image block assigns a mark, and initial markers are 0, using the markd set of institute as Label Field；By all image blocks
It is overlapped, generates three-dimensional model, calculate the median of the model, obtained result is the global mould of textile images to be measured
Plate, the template be defaulted as it is flawless, its be labeled as 0.
6. the textile flaw detection method according to claim 5 based on local optimum analysis, it is characterised in that：It is described
Calculating the method for the local energy of each image block is, for marking the image block identical with global template, counts its four neighborhood
It is interior that identical image block is marked with it, calculate the average of similarity between the image block block identical with its four neighborhoods internal labeling；
For marking the image block different from global template, its local energy is equal to the energy of the mark image block identical with global template
Maximum.
7. the textile flaw detection method based on local optimum analysis according to claim 5 or 6, it is characterised in that：
The method of the mark for changing block is the similarity factor calculated between all image blocks and global template；A decision value λ is taken to exist
(0,1) traveled through between；Change image block of the similarity factor less than λ to mark；If changing local energy after mark reduces, illustrate that mark changes
In generation, is accurate, otherwise mistake, changes back former mark.
8. the textile flaw detection method according to claim 7 based on local optimum analysis, it is characterised in that：It is described
The method that the positioning of flaw is completed according to final Label Field is flawless block to be regarded as labeled as 0 image block, labeled as 1
Image block regarded as flaw block.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710946334.1A CN107977954A (en) | 2017-10-12 | 2017-10-12 | Textile flaw detection method based on local optimum analysis |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710946334.1A CN107977954A (en) | 2017-10-12 | 2017-10-12 | Textile flaw detection method based on local optimum analysis |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107977954A true CN107977954A (en) | 2018-05-01 |
Family
ID=62012394
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710946334.1A Withdrawn CN107977954A (en) | 2017-10-12 | 2017-10-12 | Textile flaw detection method based on local optimum analysis |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107977954A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108760751A (en) * | 2018-05-25 | 2018-11-06 | 常州信息职业技术学院 | A kind of textile flaw detection method |
| CN111369480A (en) * | 2020-01-20 | 2020-07-03 | 凌云光技术集团有限责任公司 | Method and device for processing periodic texture |
| CN111929327A (en) * | 2020-09-09 | 2020-11-13 | 深兰人工智能芯片研究院（江苏）有限公司 | Cloth defect detection method and device |
| CN118247231A (en) * | 2024-03-15 | 2024-06-25 | 乐昌市恒发纺织企业有限公司 | Spinning quality visual identification system |
| CN118469134A (en) * | 2024-05-08 | 2024-08-09 | 肃宁县中原纺织有限责任公司 | Production management system of light-weight tearing-resistant multifunctional camouflage cover cloth |
-
2017
- 2017-10-12 CN CN201710946334.1A patent/CN107977954A/en not_active Withdrawn
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108760751A (en) * | 2018-05-25 | 2018-11-06 | 常州信息职业技术学院 | A kind of textile flaw detection method |
| CN111369480A (en) * | 2020-01-20 | 2020-07-03 | 凌云光技术集团有限责任公司 | Method and device for processing periodic texture |
| CN111369480B (en) * | 2020-01-20 | 2023-10-24 | 凌云光技术股份有限公司 | Method and device for processing periodic texture |
| CN111929327A (en) * | 2020-09-09 | 2020-11-13 | 深兰人工智能芯片研究院（江苏）有限公司 | Cloth defect detection method and device |
| CN118247231A (en) * | 2024-03-15 | 2024-06-25 | 乐昌市恒发纺织企业有限公司 | Spinning quality visual identification system |
| CN118247231B (en) * | 2024-03-15 | 2025-03-11 | 吴江市鑫凤织造有限公司 | Spinning quality visual identification system |
| CN118469134A (en) * | 2024-05-08 | 2024-08-09 | 肃宁县中原纺织有限责任公司 | Production management system of light-weight tearing-resistant multifunctional camouflage cover cloth |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107977954A (en) | Textile flaw detection method based on local optimum analysis | |
| CN110222681A (en) | A kind of casting defect recognition methods based on convolutional neural networks | |
| CN114066820B (en) | A fabric defect detection method based on Swin-Transformer and NAS-FPN | |
| CN110310259A (en) | A Wood Knot Defect Detection Method Based on Improved YOLOv3 Algorithm | |
| CN109859207B (en) | A defect detection method for high-density flexible substrates | |
| CN104268505B (en) | Fabric Defects Inspection automatic detecting identifier and method based on machine vision | |
| CN108257114A (en) | A kind of transmission facility defect inspection method based on deep learning | |
| CN109711474A (en) | An Algorithm for Detection of Surface Defects of Aluminum Materials Based on Deep Learning | |
| CN108428231B (en) | Multi-parameter part surface roughness learning method based on random forest | |
| CN111401419A (en) | Improved RetinaNet-based employee dressing specification detection method | |
| CN109472769A (en) | A kind of bad image defect detection method and system | |
| CN102253049B (en) | Method for accurately detecting surface quality on line in production process of band steel | |
| CN110297852B (en) | A Method of Acquiring Knowledge of Ship Painting Defects Based on PCA-Rough Sets | |
| CN118429900B (en) | Defect identification and monitoring method and system for macadimia nut production and processing | |
| CN115841491B (en) | A quality detection method for porous metal materials | |
| CN112102224B (en) | A cloth defect recognition method based on deep convolutional neural network | |
| CN108171175A (en) | A kind of deep learning sample enhancing system and its operation method | |
| CN102663422B (en) | Floor layer classification method based on color characteristic | |
| CN111815573B (en) | Coupling outer wall detection method and system based on deep learning | |
| CN114565314B (en) | A hot-rolled steel coil end surface quality control system and method based on digital twin | |
| CN115082444B (en) | Copper pipe weld defect detection method and system based on image processing | |
| CN115100206A (en) | Printing defect identification method for textile with periodic pattern | |
| CN114897865A (en) | An industrial defect detection method based on a small number of defect samples | |
| CN116205876A (en) | Unsupervised notebook appearance defect detection method based on multi-scale standardized flow | |
| Chen et al. | Real-time defect detection of TFT-LCD displays using a lightweight network architecture |
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 | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20180501 |
|
| WW01 | Invention patent application withdrawn after publication |