CN108090894A - Based on the fabric defect detection method apart from adaptation function and perception hash algorithm - Google Patents
Based on the fabric defect detection method apart from adaptation function and perception hash algorithm Download PDFInfo
- Publication number
- CN108090894A CN108090894A CN201711191878.8A CN201711191878A CN108090894A CN 108090894 A CN108090894 A CN 108090894A CN 201711191878 A CN201711191878 A CN 201711191878A CN 108090894 A CN108090894 A CN 108090894A
- Authority
- CN
- China
- Prior art keywords
- image block
- feature
- gray
- structure feature
- fabric
- 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.)
- Granted
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
- 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
- G06T7/11—Region-based segmentation
-
- 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
Abstract
The invention discloses a kind of faults based on the method detected rule fabric being combined apart from adaptation function and perception hash algorithm.First, using the minimum period that regular fabric is calculated apart from adaptation function, for obtaining image block.Repetitive unit template is built, using the structure feature for perceiving hash algorithm extraction template image block；Secondly, the structure feature and gray feature of sample to be tested epigraph block are extracted, respectively compared with the structure feature of template image block and global gray average, obtains smallest hamming distance characteristic pattern and gray average contrast characteristic figure.Finally carry out the fusion and segmentation of characteristic pattern.The result shows that the present invention has considered the structure feature of minimum period of regular fabric and gray average feature, the defect regions of fabric can be effectively extracted, realize the defect detection to fabric.
Description
Technical field
The invention belongs to the technical fields of textile images processing, and in particular to a kind of fabric defects matching detection method.
Background technology
Fabric is the basis of the consumer goods in the daily lifes such as clothes, luggage, bedding, medical cloth.Fabric Detection is
The key component of textile production quality control.At present, most of fabric detections are all to carry out vision inspection by the artificial of high cost
It looks into, but it is unreliable due to human error and eye fatigue.Automatic Visual Inspection (AVI) Appliance computer vision skill of fabric
Art, provide not only it is a kind of efficiently inexpensive and accurate method substitutes labour, but also expands detectability, with
Cover widely different textile designs, can all be applied from simplest most complicated textile design.The target of AVI is in fabric
The shape of any defect and position on fabric surface are detected and delineated during or after weaving.
For the fabric defects of pattern-free, detection algorithm can be divided mainly into statistic law, frequency domain method, modelling.Utilize statistics
Method can extract different textural characteristics from image, and background texture and fault are distinguished by the difference of statistical property
Come.Common statistical method has auto-correlation function, co-occurrence matrix, mathematical morphology, fractal dimension etc..
In frequency domain method, mainly image is transformed under some frequency domain, is handled to image.Mainly there is Fourier
Conversion, wavelet transformation and Gabor filter etc..The calculating of this kind of method is relative complex, and the optimization of parameter influences result
It is bigger.
The texture of general fabrics had not only included the ingredient of regularity, but also had the ingredient changed at random, and modelling is suitable for surface
The fabric that texture changes at random, and statistic law and Spectrum Method are difficult to accomplish.It is typical to check that algorithm has markov random file, from
Regression model.
Most of algorithm being suggested is all based on the pattern-free Fabric Defect Detection of simple textures.But to figure
For the fabric of case, fabric different texture and pattern and the similitude of fault and background all can bring great challenge to detection.
The content of the invention
In order to solve the technical issues of above-mentioned, it is based on the present invention provides one kind apart from adaptation function and perceives hash algorithm
Fabric defect detection method, realize the effective detection and positioning to regular fabric defects, and with higher inspection precision.
Technical solution provided by the invention is：
A kind of fabric defect detection method based on apart from adaptation function and perception hash algorithm, its step are as follows：
Step 1：Using minimum Texture-period is extracted from normal rule image apart from adaptation function；
Step 2：The template of repetition period is constructed, using the structure feature for perceiving hash algorithm extraction template image block；
Step 3：Testing image block is obtained from sample to be tested, extracts the structure feature and gray average of testing image block
Feature；
Step 4：For the structure feature of each testing image block, compared one by one with the structure feature of template image block, it is raw
Into smallest hamming distance characteristic pattern H；
Step 5：For the gray feature of each testing image block, compared with the gray feature of sample to be tested, generate gray scale
Average contrast characteristic schemes D；
Step 6：Fusion feature figure is split final characteristic pattern using maximum variance between clusters, orients fault
Region.
It is described to use the method for extracting minimum Texture-period from normal rule image apart from adaptation function：With two-dimentional variable f
(x, y) represents textile image gray value, calculates two-dimensional distance adaptation function：
WithIn formula, M, N are respectively the wide and high of the image, and x, y distinguish
For the row and column of pixel in the image；The cycle of p representative functions, when the line direction or the week of column direction that p is function f (x, y)
During the phase, the difference of f (x, y) and f (x, y+p) or f (x+p, y) is minimum, remembers width high respectively a, b of minimum period, subsequent extracted
Tile size be minimum period size.
The construction repeats template and refers to：From normal rule textile image, the positive routine of wide, a height of 2a and 2b is obtained
Then image template.
The flow of the perception hash algorithm is：1) image block is narrowed down to 8 × 8 size；2) scheme after calculating compression
As the average gray of block；3) gray scale of compared pixels：By the gray scale of each pixel, compared with average value, it is more than or waits
In average value, 1 is denoted as；Less than average value, 0 is denoted as；4) cryptographic Hash is calculated：It by the comparative result of previous step, combines, just
The integer of one 64 is constituted, ensures that all image blocks all use similary order.
The described method using the structure feature for perceiving hash algorithm extraction image block is：With the minimum period calculated
A × b is window, is slided in template with fixed step size and obtains image block, image is calculated using the perception hash algorithm
The cryptographic Hash of block is as structure feature.
Testing image block is obtained on the slave sample to be tested, structure feature and the gray average for extracting testing image block are special
The method of sign is：Using the minimum period a × b calculated as window, slided on sample to be tested with fixed step size and obtain image block,
The cryptographic Hash for calculating image block using perception hash algorithm calculates the average gray value conduct of image block as structure feature
Gray feature.
The structure feature for each testing image block, compares one by one with the structure feature of template image block, and generation is most
The method of small Hamming distance characteristic pattern H is：On sample to be tested with the minimum period obtain testing image block, then by testing image block with
The structure feature of template image block compares one by one, takes the minimum value of Hamming distance as a comparison as a result, being denoted as smallest hamming distance,
During generating characteristic pattern, a pixel of a testing image block representative feature figure, smallest hamming distance is characterized a little
Pixel value, wherein, Hamming distance dh(x, y) represents the similarity between two image cryptographic Hash x, y, can be by two Hash
Value carries out XOR operation, and the number that statistical result is 1 obtains Hamming distance, and formula is
The gray feature for each testing image block compares with the gray feature of sample to be tested, generates gray scale
Average contrast characteristic scheme D method be：Testing image block is obtained with the minimum period on sample to be tested, by the ash of testing image block
The average gray of degree average value and sample to be tested, which makes the difference, to take absolute value as a comparison again as a result, generating the process of characteristic pattern
In, a pixel of a testing image block representative feature figure, comparing result is characterized pixel value a little.
The fusion feature figure is split characteristic pattern using maximum variance between clusters, the method for orienting defect regions
It is：Characteristic pattern is subjected to threshold value division according to formula firstWith
According still further to formula M (i, j)=(H (i, j)+D (i, j))2Fusion feature figure finally carries out binaryzation using maximum variance between clusters,
Wherein θ1For the threshold value of structure feature, θ2For the threshold value of gray feature, M (i, j) is the pixel of characteristic pattern after fusion.
The advantage of the invention is that：
1) structure feature of Texture-period is extracted using perception hash algorithm, effectively illustrates the structure feature of fabric,
Calculating speed is very fast.
2) consider the structure feature and gray feature of the image block of minimum period, complete regular fabric defects
Detection.
Description of the drawings
Fig. 1 is the Fabric Defect Detection flow chart of the present invention.
Fig. 2 is that the present invention perceives hash algorithm flow chart.
Fig. 3 is the image to be split with the minimum period.
Fig. 4 is the minimum period and template to be partitioned into.
Fig. 5 is defect image (the first row)；Go out the characteristic pattern of smallest hamming distance using perception hash algorithm comparing calculation
(the second row)；Go out the characteristic pattern (the third line) of intensity contrast using gray average comparing calculation；The fusion feature of this method generation
Scheme (fourth line)；Segmentation result (fifth line).
Specific embodiment
The present invention is described in further detail below in conjunction with the accompanying drawings.
This patent proposes a kind of based on apart from adaptation function and perceiving the method detected rule that is combined of hash algorithm and knit
The fault of object.First, using the minimum period that regular fabric is calculated apart from adaptation function, for obtaining image block.Structure weight
Multiple unit template, using the structure feature for perceiving hash algorithm extraction template image block；Secondly, sample to be tested epigraph block is extracted
Structure feature and gray feature, respectively compared with the structure feature of template image block and global gray average, obtain most
Small Hamming distance characteristic pattern and gray average contrast characteristic figure.Finally carry out the fusion and segmentation of characteristic pattern.The result shows that this
Invention has considered the structure feature of minimum period of regular fabric and gray average feature, can effectively extract fabric
Defect regions, realize the defect detection to fabric.
It is as shown in Figure 1, a kind of based on the method detected rule fabric being combined apart from adaptation function and perception hash algorithm
Fault step it is as follows：
Step 1：Using minimum Texture-period is extracted from normal rule image apart from adaptation function；
It is using the method for extracting minimum Texture-period from normal rule image apart from adaptation function：With two-dimentional variable f (x,
Y) represent textile image gray value, calculate two dimension apart from adaptation function：
With
In formula, M, N are respectively the wide and high of the image, and M=N=256 in this example, x, y is respectively pixel in the image
Row and column；The cycle of p representative functions, when the cycle of line direction or column direction that p is function f (x, y), f (x, y) and f (x,
Y+p) or the difference of f (x+p, y) is minimum.Remember width high respectively a, b of minimum period, the tile size of subsequent extracted is most
Minor cycle size.If Fig. 3 is with the textile image of minimum period segmentation.
Step 2：The template of repetition period is constructed, using the structure feature for perceiving hash algorithm extraction template image block；
Construction repeats template and refers to：From normal rule textile image, the normal rule image of wide, a height of 2a and 2b is obtained
Template, as shown in Figure 4.
Perceiving the flow of hash algorithm is, as shown in Fig. 2 flows：1) image block is narrowed down to 8 × 8 size；2) calculate
The average gray of compressed images block；3) gray scale of compared pixels：By the gray scale of each pixel, compared with average value.
More than or equal to average value, 1 is denoted as；Less than average value, 0 is denoted as；4) cryptographic Hash is calculated：By the comparative result of previous step, combination
Together, the integer of one 64 is just constituted, ensures that all image blocks all use similary order.
Referred to using the structure feature for perceiving hash algorithm extraction image block：Using the minimum period a × b calculated as window
Mouthful, it is slided in template with fixed step size and obtains image block, the Hash of image block is calculated using the perception hash algorithm
Value is as structure feature.In this example, fixed sliding step is 2.
Step 3：Testing image block is obtained from sample to be tested, extracts the structure feature and gray average of testing image block
Feature；
Testing image block is obtained from sample to be tested, extracts the structure feature of testing image block and the side of gray average feature
Method is：Using the minimum period a × b calculated as window, slided on sample to be tested with fixed step size and obtain image block, using sense
Know that hash algorithm calculates the cryptographic Hash of image block as structure feature, the average gray value for calculating image block is special as gray scale
Sign.
Step 4：For the structure feature of each testing image block, compared one by one with the structure feature of template image block, it is raw
Into smallest hamming distance characteristic pattern H；
It for the structure feature of each testing image block, is compared one by one with the structure feature of template image block, generation is minimum
The method of Hamming distance characteristic pattern H is：On sample to be tested with the minimum period obtain testing image block, then by testing image block with
The structure feature of template image block compares one by one, takes the minimum value of Hamming distance as smallest hamming distance.In generation characteristic pattern
During, a pixel of a testing image block representative feature figure, smallest hamming distance is characterized pixel value a little.Its
In, Hamming distance dh(x, y) represents the similarity between two image cryptographic Hash x, y.It can be by being carried out to two cryptographic Hash
XOR operation, and the number that statistical result is 1 obtains Hamming distance, formula is
Step 5：For the gray feature of each testing image block, compared with the gray feature of sample to be tested, generate gray scale
Average contrast characteristic schemes D；
For the gray feature of each testing image block, compared with the gray feature of sample to be tested, generate gray average pair
Method than characteristic pattern D is：Testing image block is obtained with the minimum period on sample to be tested, the gray scale of testing image block is averaged
Value and the average gray of sample to be tested make the difference the result as a comparison that takes absolute value again.During characteristic pattern is generated, one
One pixel of testing image block representative feature figure, comparing result are characterized pixel value a little.
Step 6：Fusion feature figure is split final characteristic pattern using maximum variance between clusters, orients fault
Region.
Fusion feature figure is split characteristic pattern using maximum variance between clusters, and orienting the method for defect regions is：
Characteristic pattern is subjected to threshold value division according to formula firstWith
According still further to formula M (i, j)=(H (i, j)+D (i, j))2Fusion feature figure.Finally two-value is carried out using maximum variance between clusters
Change.Wherein θ1For the threshold value of structure feature, θ2For the threshold value of gray feature.M (i, j) is the pixel of characteristic pattern after fusion.
Embodiment：The common fabric defect image of several classes (broken hole, disconnected warp, oil stain, cut etc.) is selected from textile image storehouse,
Picture size is 256pixel X256pixel.Parts of images is selected, such as Fig. 5 (the first row).It is obtained using inventive algorithm
Structure feature figure and gray feature figure, such as Fig. 5 (the second row, the third line).Fusion feature figure and defect segmentation image are finally obtained,
Wherein, θ is taken1=6, θ2=8, as a result such as Fig. 5 (fourth line, fifth line), obtain final result (the 6th row), it can be seen that fault
It is detected.
Claims (9)
1. based on the fabric defect detection method apart from adaptation function and perception hash algorithm, which is characterized in that its step are as follows：
Step 1：Using minimum Texture-period is extracted from normal rule image apart from adaptation function；
Step 2：The template of repetition period is constructed, using the structure feature for perceiving hash algorithm extraction template image block；
Step 3：Testing image block is obtained from sample to be tested, extracts the structure feature of testing image block and gray average feature；
Step 4：It for the structure feature of each testing image block, is compared one by one with the structure feature of template image block, generation is most
Small Hamming distance characteristic pattern H；
Step 5：For the gray feature of each testing image block, compared with the gray feature of sample to be tested, generate gray average
Contrast characteristic schemes D；
Step 6：Fusion feature figure is split final characteristic pattern using maximum variance between clusters, orients fault area
Domain.
2. fabric defect detection method according to claim 1, which is characterized in that described to use apart from adaptation function from just
The method that minimum Texture-period is extracted in normal regular image：Textile image gray value is represented with two-dimentional variable f (x, y), calculates two
Dimension is apart from adaptation function：
WithIn formula, M, N are respectively the wide and high of the image,
X, y are respectively the row and column of pixel in the image；The cycle of p representative functions, when the line direction or row side that p is function f (x, y)
To cycle when, the difference of f (x, y) and f (x, y+p) or f (x+p, y) are minimum, remember the width of minimum period it is high be respectively a, b, after
The tile size of continuous extraction is minimum period size.
3. fabric defect detection method according to claim 1, which is characterized in that the construction repeats template and refers to：
From normal rule textile image, the normal rule image template of wide, a height of 2a and 2b is obtained.
4. fabric defect detection method according to claim 1, which is characterized in that the flow of the perception hash algorithm
It is：1) image block is narrowed down to 8 × 8 size；2) average gray of compressed images block is calculated；3) ash of compared pixels
Degree：By the gray scale of each pixel, compared with average value, more than or equal to average value, 1 is denoted as；Less than average value, 0 is denoted as；
4) cryptographic Hash is calculated：It by the comparative result of previous step, combines, just constitutes the integer of one 64, ensure all figures
As block all uses similary order.
5. fabric defect detection method according to claim 1, which is characterized in that described is carried using perception hash algorithm
Taking the method for the structure feature of image block is：Using the minimum period a × b calculated as window, slided in template with fixed step size
It is dynamic to obtain image block, the cryptographic Hash of image block is calculated as structure feature using the perception hash algorithm.
6. fabric defect detection method according to claim 1, which is characterized in that described obtained from sample to be tested is treated
Altimetric image block, the method for the structure feature and gray average feature of extracting testing image block are：With calculate minimum period a ×
B is window, is slided on sample to be tested with fixed step size and obtains image block, and the Kazakhstan of image block is calculated using perception hash algorithm
Uncommon value calculates the average gray value of image block as gray feature as structure feature.
7. fabric defect detection method according to claim 1, which is characterized in that described is directed to each testing image block
Structure feature, compared one by one with the structure feature of template image block, the method for generation smallest hamming distance characteristic pattern H is：
Testing image block is obtained with the minimum period on sample to be tested, then the structure feature one of testing image block and template image block is a pair of
Than, the minimum value of Hamming distance is taken as a comparison as a result, being denoted as smallest hamming distance, during characteristic pattern is generated, one
One pixel of testing image block representative feature figure, smallest hamming distance are characterized pixel value a little, wherein, Hamming distance dh
(x, y) represents the similarity between two image cryptographic Hash x, y, can be by carrying out XOR operation to two cryptographic Hash, and unites
The number that meter result is 1 obtains Hamming distance, and formula is
8. fabric defect detection method according to claim 1, which is characterized in that described is directed to each testing image block
Gray feature, compared with the gray feature of sample to be tested, the method that generation gray average contrast characteristic schemes D is：Treating test sample
Testing image block is obtained with the minimum period on this, the average gray of the average gray of testing image block and sample to be tested is done
Difference is taken absolute value again as a comparison as a result, during characteristic pattern is generated, one of a testing image block representative feature figure
Pixel, comparing result are characterized pixel value a little.
9. fabric defect detection method according to claim 1, which is characterized in that the fusion feature figure, using most
Big Ostu method is split characteristic pattern, and orienting the method for defect regions is：Characteristic pattern is carried out according to formula first
Threshold value dividesWithAccording still further to formula M (i,
J)=(H (i, j)+D (i, j))2Fusion feature figure finally carries out binaryzation, wherein θ using maximum variance between clusters1It is special for structure
The threshold value of sign, θ2For the threshold value of gray feature, M (i, j) is the pixel of characteristic pattern after fusion.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711191878.8A CN108090894B (en) | 2017-11-24 | 2017-11-24 | Fabric defect detection method based on distance matching function and perceptual hash algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711191878.8A CN108090894B (en) | 2017-11-24 | 2017-11-24 | Fabric defect detection method based on distance matching function and perceptual hash algorithm |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108090894A true CN108090894A (en) | 2018-05-29 |
| CN108090894B CN108090894B (en) | 2020-09-08 |
Family
ID=62172779
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711191878.8A Active CN108090894B (en) | 2017-11-24 | 2017-11-24 | Fabric defect detection method based on distance matching function and perceptual hash algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108090894B (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109344676A (en) * | 2018-11-22 | 2019-02-15 | 福州图腾易讯信息技术有限公司 | A kind of automatic sensing triggering method and its system based on hash algorithm |
| CN109344273A (en) * | 2018-07-25 | 2019-02-15 | 中兴通讯股份有限公司 | A kind of method, apparatus and mobile terminal of wallpaper management |
| CN109598726A (en) * | 2018-10-26 | 2019-04-09 | 哈尔滨理工大学 | A kind of adapting to image target area dividing method based on SLIC |
| CN109685114A (en) * | 2018-11-29 | 2019-04-26 | 昆明理工大学 | It is a kind of based on prescreening method determination method similar with the image of PHash |
| CN109685112A (en) * | 2018-11-29 | 2019-04-26 | 昆明理工大学 | It is a kind of based on color difference algorithm determination method similar with the image of DHash |
| CN109961432A (en) * | 2019-03-27 | 2019-07-02 | 广东工业大学 | A kind of detection method and system of filter cloth breakage |
| CN110111292A (en) * | 2019-04-30 | 2019-08-09 | 淮阴师范学院 | A kind of infrared and visible light image fusion method |
| CN110120029A (en) * | 2019-04-30 | 2019-08-13 | 淮阴师范学院 | Image interfusion method based on perceptual hash algorithm |
| CN111709915A (en) * | 2020-05-28 | 2020-09-25 | 拉萨经济技术开发区美第意户外用品有限公司 | Automatic detection method and system for quick-drying fabric defects |
| CN111814807A (en) * | 2020-07-16 | 2020-10-23 | 北京字节跳动网络技术有限公司 | Method, apparatus, electronic device, and computer-readable medium for processing image |
| CN114419039A (en) * | 2022-03-28 | 2022-04-29 | 武汉市融科优品装饰材料有限公司 | Decorative wallpaper defect detection method based on template matching |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103234976A (en) * | 2013-04-03 | 2013-08-07 | 江南大学 | Warp knitting machine cloth flaw on-line visual inspection method based on Gabor transformation |
| CN103778624A (en) * | 2013-12-20 | 2014-05-07 | 中原工学院 | Fabric defect detection method based on optical threshold segmentation |
| CN104778692A (en) * | 2015-04-09 | 2015-07-15 | 中原工学院 | Fabric defect detection method based on sparse representation coefficient optimization |
| CN105261003A (en) * | 2015-09-10 | 2016-01-20 | 西安工程大学 | Defect point detection method on basis of self structure of fabric |
| CN106780464A (en) * | 2016-12-15 | 2017-05-31 | 东华大学 | A kind of fabric defect detection method based on improvement Threshold segmentation |
| CN107085844A (en) * | 2017-03-14 | 2017-08-22 | 西安工程大学 | The fabric defects detection method of picture breakdown algorithm based on rarefaction representation |
| CN107240086A (en) * | 2016-03-25 | 2017-10-10 | 南通大学 | A kind of fabric defects detection method based on integration nomography |
-
2017
- 2017-11-24 CN CN201711191878.8A patent/CN108090894B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103234976A (en) * | 2013-04-03 | 2013-08-07 | 江南大学 | Warp knitting machine cloth flaw on-line visual inspection method based on Gabor transformation |
| CN103778624A (en) * | 2013-12-20 | 2014-05-07 | 中原工学院 | Fabric defect detection method based on optical threshold segmentation |
| CN104778692A (en) * | 2015-04-09 | 2015-07-15 | 中原工学院 | Fabric defect detection method based on sparse representation coefficient optimization |
| CN105261003A (en) * | 2015-09-10 | 2016-01-20 | 西安工程大学 | Defect point detection method on basis of self structure of fabric |
| CN107240086A (en) * | 2016-03-25 | 2017-10-10 | 南通大学 | A kind of fabric defects detection method based on integration nomography |
| CN106780464A (en) * | 2016-12-15 | 2017-05-31 | 东华大学 | A kind of fabric defect detection method based on improvement Threshold segmentation |
| CN107085844A (en) * | 2017-03-14 | 2017-08-22 | 西安工程大学 | The fabric defects detection method of picture breakdown algorithm based on rarefaction representation |
Non-Patent Citations (4)
| Title |
|---|
| IBRAHIM CEM BAYKAL ET AL: "On the Use of Hash Functions as Preprocessing Algorithms to Detect Defects on Repeating Definite Textures", 《MACHINE VISION AND APPLICATIONS》 * |
| JING JUNFENG ET AL: "Patterned fabric defect detection via convolutional matching pursuit dual-dictionary", 《OPTICAL ENGINEERING》 * |
| 刘洲峰: "基于局部统计与整体显著性的织物疵点检测算法", 《纺织学报》 * |
| 李春雷: "应用方向梯度直方图和低秩分解的织物疵点检测算法", 《纺织学报》 * |
Cited By (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109344273B (en) * | 2018-07-25 | 2021-06-15 | 中兴通讯股份有限公司 | Wallpaper management method and device and mobile terminal |
| CN109344273A (en) * | 2018-07-25 | 2019-02-15 | 中兴通讯股份有限公司 | A kind of method, apparatus and mobile terminal of wallpaper management |
| US11462002B2 (en) | 2018-07-25 | 2022-10-04 | Zte Corporation | Wallpaper management method, apparatus, mobile terminal, and storage medium |
| CN109598726A (en) * | 2018-10-26 | 2019-04-09 | 哈尔滨理工大学 | A kind of adapting to image target area dividing method based on SLIC |
| CN109344676A (en) * | 2018-11-22 | 2019-02-15 | 福州图腾易讯信息技术有限公司 | A kind of automatic sensing triggering method and its system based on hash algorithm |
| CN109344676B (en) * | 2018-11-22 | 2021-09-24 | 福州图腾易讯信息技术有限公司 | Automatic induction triggering method and system based on Hash algorithm |
| CN109685114B (en) * | 2018-11-29 | 2022-04-15 | 昆明理工大学 | Image similarity judgment method based on pre-screening method and PHash |
| CN109685112A (en) * | 2018-11-29 | 2019-04-26 | 昆明理工大学 | It is a kind of based on color difference algorithm determination method similar with the image of DHash |
| CN109685114A (en) * | 2018-11-29 | 2019-04-26 | 昆明理工大学 | It is a kind of based on prescreening method determination method similar with the image of PHash |
| CN109961432A (en) * | 2019-03-27 | 2019-07-02 | 广东工业大学 | A kind of detection method and system of filter cloth breakage |
| CN110120029A (en) * | 2019-04-30 | 2019-08-13 | 淮阴师范学院 | Image interfusion method based on perceptual hash algorithm |
| CN110111292A (en) * | 2019-04-30 | 2019-08-09 | 淮阴师范学院 | A kind of infrared and visible light image fusion method |
| CN110120029B (en) * | 2019-04-30 | 2023-06-23 | 淮阴师范学院 | Image fusion method based on perceptual hash algorithm |
| CN110111292B (en) * | 2019-04-30 | 2023-07-21 | 淮阴师范学院 | Infrared and visible light image fusion method |
| CN111709915A (en) * | 2020-05-28 | 2020-09-25 | 拉萨经济技术开发区美第意户外用品有限公司 | Automatic detection method and system for quick-drying fabric defects |
| CN111814807A (en) * | 2020-07-16 | 2020-10-23 | 北京字节跳动网络技术有限公司 | Method, apparatus, electronic device, and computer-readable medium for processing image |
| CN111814807B (en) * | 2020-07-16 | 2023-10-24 | 抖音视界有限公司 | Method, apparatus, electronic device, and computer-readable medium for processing image |
| CN114419039A (en) * | 2022-03-28 | 2022-04-29 | 武汉市融科优品装饰材料有限公司 | Decorative wallpaper defect detection method based on template matching |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108090894B (en) | 2020-09-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108090894A (en) | Based on the fabric defect detection method apart from adaptation function and perception hash algorithm | |
| Çelik et al. | Development of a machine vision system: real-time fabric defect detection and classification with neural networks | |
| Tsang et al. | Fabric inspection based on the Elo rating method | |
| Zhang et al. | Fabric defect detection using salience metric for color dissimilarity and positional aggregation | |
| CN103604809B (en) | A kind of online visible detection method of pattern cloth flaw | |
| CN105277567B (en) | A kind of fabric defects detection method | |
| CN106373124B (en) | Industrial products surface defect visible detection method based on gray level co-occurrence matrixes and RANSAC | |
| CN104198498B (en) | Fabric Defects Inspection detection method based on adaptive orthogonal wavelet conversion and device | |
| CN103903275B (en) | Method for improving image segmentation effects by using wavelet fusion algorithm | |
| CN103425984B (en) | A kind of detect the method and apparatus of regular polygonal seal in bill | |
| Zhang et al. | A review of fabric identification based on image analysis technology | |
| CN103234969A (en) | Method for measuring fabric weft density based on machine vision | |
| Schneider et al. | Blind weave detection for woven fabrics | |
| Pawening et al. | Classification of textile image using support vector machine with textural feature | |
| CN108133473A (en) | Warp knitted jacquard fabric defect detection method based on Gabor filtering and deep neural network | |
| CN101976442B (en) | Method for extracting fractal profile for representing fabric texture and Sobel operator filtering detail mixed characteristic vector | |
| CN107966444A (en) | Textile flaw detection method based on template | |
| CN113160123A (en) | Leather defect detection method, system and device based on YOLOv5 | |
| NU | Automatic detection of texture defects using texture-periodicity and Gabor wavelets | |
| CN102289677A (en) | Method for analyzing image based on principal component analysis and method applicable to detection of defects of fabric | |
| Chandra et al. | Neural network trained morphological processing for the detection of defects in woven fabric | |
| CN101976441B (en) | Method for extracting Sobel operator filtering profile for representing fabric texture and fractal detail mixed characteristic vector | |
| Perng et al. | Automatic surface inspection for directional textures using nonnegative matrix factorization | |
| Liu et al. | Fabric defect detection algorithm based on convolution neural network and low-rank representation | |
| Jiang et al. | Automatic identification of jacquard warp-knitted fabric patterns based on the wavelet transform |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20210311 Address after: 211111 No. 12, Mazhou East Road, Mau Ling Street, Jiangning District, Nanjing, Jiangsu Patentee after: JIANGSU ZHIJU INTELLECTUAL PROPERTY SERVICE Co.,Ltd. Address before: 310018 No. 928 street, Baiyang street, Xiasha Higher Education Park, Jianggan District, Hangzhou, Zhejiang, China. 928 Patentee before: Zhejiang University of Technology |
|
| TR01 | Transfer of patent right |