CN109858485A - A kind of fabric defects detection method based on LBP and GLCM - Google Patents
A kind of fabric defects detection method based on LBP and GLCM Download PDFInfo
- Publication number
- CN109858485A CN109858485A CN201910074036.7A CN201910074036A CN109858485A CN 109858485 A CN109858485 A CN 109858485A CN 201910074036 A CN201910074036 A CN 201910074036A CN 109858485 A CN109858485 A CN 109858485A
- Authority
- CN
- China
- Prior art keywords
- glcm
- lbp
- feature
- fabric
- image
- 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
Landscapes
- Treatment Of Fiber Materials (AREA)
- Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
Abstract
The present invention relates to a kind of fabric defects detection methods based on LBP and GLCM, include: that template characteristic extracts the stage: for the flawless standard picture of fabric, the standard feature threshold value and decision condition that detect same kind fabric are determined with the GLCM feature extracting method based on LBP；The Fabric Detection stage: the characteristic value of image to be detected is extracted, and extracts the standard feature threshold value that the stage determines with template characteristic and is compared, the Defect Detection to image to be detected is realized according to comparison result and decision condition.The present invention can fast implement the automatic detection and positioning of fabric defects.
Description
Technical field
The present invention relates to fabric defects detection technique fields, more particularly to a kind of fabric defects based on LBP and GLCM
Detection method.
Background technique
Fabric defects detection is the important link of textile printing and dyeing quality monitoring, and traditional artificial detection method is not only time-consuming to be consumed
Power, and due to the experience of inspector and visual fatigue etc., it is difficult to guarantee detection quality.In this context, one is needed
Kind knits flaw detection method, and artificial offline inspection can be replaced to realize automatic detection and positioning to fabric defects.
Summary of the invention
Technical problem to be solved by the invention is to provide a kind of fabric defects detection method based on LBP and GLCM, energy
Enough fast implement the automatic detection and positioning of fabric defects.
The technical solution adopted by the present invention to solve the technical problems is: providing a kind of fabric flaw based on LBP and GLCM
Defect detection method, comprising:
(1) template characteristic extracts the stage: being directed to the flawless standard picture of fabric, mentions with the GLCM feature based on LBP
Method is taken to determine to detect the standard feature threshold value of same kind fabric and decision condition；
(2) the Fabric Detection stage: the characteristic value of image to be detected is extracted, and extracts stage determining standard with template characteristic
Characteristic threshold value is compared, and the Defect Detection to image to be detected is realized according to comparison result and decision condition.
The step (1) includes following sub-step:
(11) indefectible fabric standard picture is chosen as template, image is pre-processed, and completes the filtering behaviour of image
Make；
(12) image is not divided into overlappingly to N number of Wd×WdThe detection window of size uses each detection window
GLCM feature extracting method based on LBP extracts feature vector；
(13) after N number of detection window, which all extracts, to be finished, the respective intermediate value of single feature value in N group feature vector is acquired
Obtain feature mean vectors；
(14) Dynamic gene is chosen respectively to each feature；
(15) using the product of the Dynamic gene of selection and feature mean vectors as standard feature threshold value, and it is directed to each spy
Value indicative determines judgement of the decision condition for flaw.
Pretreatment in the step (11) specifically: preliminary treatment is carried out to textile image, including mask size is 3*3
Gaussian filtering, gray proces and mask size be 7*7 gaussian filtering.
The GLCM feature extracting method based on LBP in the step (12) extracts feature vector specifically:
(a) it usesImage gray levels are reduced to 0-P+1 by operator, whereinOperator is that have rotation not
The local binary pattern operator of denaturation and " unification " mode, P indicate pixel number in center pixel vertex neighborhood, and R indicates neighborhood
Radius；
(b) gray level co-occurrence matrixes of 0 °, 45 °, 90 ° and 135 ° four direction are generated on this basis；
(c) average value of four gray level co-occurrence matrixes is calculated as mixing gray level co-occurrence matrixes；
(d) characteristic value of mixing gray level co-occurrence matrixes is extracted as feature vector.
Feature vector in the step (d) includes: energyEntropyContrastUnfavourable balance away fromAutocorrelationInconsistency
Cluster shadeWith significant cluster
Wherein,P (i, j) indicates that gray value is respectively the pixel of i and j in ash
The entry values in co-occurrence matrix are spent, L indicates image gray levels.
The step (2) includes following sub-step:
(21) textile image to be detected is pre-processed, completes the filtering operation of image；
(22) detection window identical with template characteristic extraction stage size is chosen, according to fixed step-length m, first from left-hand
It is right to slide detection from overlapping downwards again；
(23) feature vector of detection window is extracted using the GLCM feature extracting method based on LBP；
(24) feature vector and standard feature threshold value that will test window are compared, and carry out flaw window according to decision condition
The calibration of mouth.
Pretreatment mode in the step (21) is identical with the pretreatment mode in template characteristic extraction stage；The step
(23) the GLCM feature extracting method in based on LBP is identical with the extracting method in template characteristic extraction stage.
Beneficial effect
Due to the adoption of the above technical solution, compared with prior art, the present invention having the following advantages that and actively imitating
Fruit: the present invention is divided into template characteristic extraction and two stages of Fabric Detection, extracts the stage in template characteristic, determines to detect similar
The standard feature threshold value and decision condition of fabric；In the Fabric Detection stage, the texture eigenvalue that will be extracted in image to be detected
It is compared with standard feature threshold value, realizes identifying and positioning for flaw according to decision condition.This method computation complexity is lower,
It is able to achieve being dynamically determined for threshold value, to the well adapting to property of fabric of tiny texture and coarse grain.
Detailed description of the invention
Fig. 1 is the flow chart of step 1 template characteristic extraction process in embodiment of the present invention；
Fig. 2 is the flow chart of step 2 fabric defects detection process in embodiment of the present invention；
Fig. 3 is the flow chart of image preprocessing process in embodiment of the present invention；
Fig. 4 is the flow chart of the GLCM feature extracting method in embodiment of the present invention based on LBP.
Specific embodiment
Present invention will be further explained below with reference to specific examples.It should be understood that these embodiments are merely to illustrate the present invention
Rather than it limits the scope of the invention.In addition, it should also be understood that, after reading the content taught by the present invention, those skilled in the art
Member can make various changes or modifications the present invention, and such equivalent forms equally fall within the application the appended claims and limited
Range.
Embodiments of the present invention are related to a kind of fabric defects detection method based on LBP and GLCM, and this method is divided into mould
Plate features extract and two stages of Fabric Detection.The stage is extracted in template characteristic, determines the standard feature for detecting same kind fabric
Threshold value and decision condition；In the Fabric Detection stage, by the texture eigenvalue extracted in image to be detected and standard feature threshold value
It is compared, realizes identifying and positioning for flaw according to decision condition.Specifically includes the following steps:
Step 1: template characteristic is extracted: being directed to the indefectible standard picture of fabric, is determined that the standard for detecting same kind fabric is special
Levy threshold value and decision condition.As shown in Figure 1, the main sub-processes of the step include:
1.1 choose indefectible fabric standard picture as template, pre-process to image, complete the filtering behaviour of image
Make, is detailed in image preprocessing process；
1.2 are not divided into image overlappingly N number of Wd×WdThe detection window of size, for each detection window, using base
Feature vector is extracted in the GLCM feature extracting method of LBP；
1.3 after N number of detection window is all extracted and is finished, and acquires the respective intermediate value of single feature value in N group feature vector
Obtain feature mean vectors；
1.4 pairs of each features choose reasonable Dynamic gene respectively；
1.5 choose the product of Dynamic gene and feature mean vectors as standard feature threshold value, and are directed to each characteristic value
Determine judgement of the decision condition for flaw.
Step 2: fabric defects detection: the characteristic value and template characteristic for extracting image to be detected extract the standard determined in the stage
Characteristic threshold value is compared, and the Defect Detection to image to be detected is realized according to comparison result and decision condition.As shown in Fig. 2,
The main sub-processes of the step include:
2.1 pairs of textile images to be detected pre-process, and complete the filtering operation of image, are detailed in image preprocessing process；
2.2 choose detection window identical with template characteristic extraction stage size, according to fixed step-length m, from left to right
Overlapping sliding detection from up to down；
2.3 extract the feature vector of detection window using the GLCM feature extracting method based on LBP；
2.4 will test the feature vector of window and standard feature threshold value is compared, and carry out flaw window according to decision condition
The calibration of mouth；
The detection of 2.5 image all areas finishes, and obtains testing result.
Identical image pre-processing method can be used in step 1 and step 2, process is as shown in figure 3, it is used to complete
The filtering of image and gray processing work, main contents are as follows:
Preliminary treatment is carried out to textile image, the gaussian filtering for being 3*3 including mask size, gray proces, mask size
For the gaussian filtering of 7*7.
The GLCM feature extracting method based on LBP in present embodiment combine LBP operator description local feature and
The global characteristics of GLCM description, extract the texture eigenvalue in fabric figure, and characteristic value has gray scale invariance and invariable rotary
Property and lower computation complexity.Main contents are as follows:
1) it usesImage gray levels are reduced to 0-P+1 by operator；Wherein,Operator is with invariable rotary
Property and " unification " mode local binary pattern operator, P indicate center pixel vertex neighborhood in pixel number, R indicate neighborhood partly
Diameter.
2) GLCM of 0 °, 45 °, 90 ° and 135 ° four direction is generated on this basis；
3) average value of four GLCM is further sought as mixing GLCM；
4) characteristic value of mixing GLCM is extracted as feature vector, wherein characteristic value is successively are as follows:
(a) energy
(b) entropy
(c) contrast
(d) unfavourable balance away from
(e) autocorrelation
(f) inconsistency
(g) shade is clustered
(h) significant cluster
Wherein,P (i, j) indicates that gray value is respectively the picture of i and j
Element indicates image gray levels to the entry values in gray level co-occurrence matrixes, L.
It is not difficult to find that the present invention is divided into template characteristic extraction and two stages of Fabric Detection, the stage is extracted in template characteristic,
Determine to detect the standard feature threshold value and decision condition of same kind fabric；In the Fabric Detection stage, will be extracted in image to be detected
To texture eigenvalue be compared with standard feature threshold value, realize that flaw identifies and positions according to decision condition.This method
Computation complexity is lower, is able to achieve being dynamically determined for threshold value, well adapts to the fabric of tiny texture and coarse grain
Property.
Claims (7)
1. a kind of fabric defects detection method based on LBP and GLCM characterized by comprising
(1) template characteristic extracts the stage: the flawless standard picture of fabric is directed to, with the feature extraction side GLCM based on LBP
Method is determined to detect the standard feature threshold value of same kind fabric and decision condition；
(2) the Fabric Detection stage: the characteristic value of image to be detected is extracted, and extracts stage determining standard feature with template characteristic
Threshold value is compared, and the Defect Detection to image to be detected is realized according to comparison result and decision condition.
2. the fabric defects detection method according to claim 1 based on LBP and GLCM, which is characterized in that the step
(1) include following sub-step:
(11) indefectible fabric standard picture is chosen as template, and image is pre-processed, the filtering operation of image is completed；
(12) image is not divided into overlappingly to N number of Wd×WdThe detection window of size, for each detection window, using being based on
The GLCM feature extracting method of LBP extracts feature vector；
(13) it after N number of detection window, which all extracts, to be finished, acquires the respective intermediate value of single feature value in N group feature vector and obtains
Feature mean vectors；
(14) Dynamic gene is chosen respectively to each feature；
(15) using the product of the Dynamic gene of selection and feature mean vectors as standard feature threshold value, and it is directed to each characteristic value
Determine judgement of the decision condition for flaw.
3. the fabric defects detection method according to claim 2 based on LBP and GLCM, which is characterized in that the step
(11) pretreatment in specifically: preliminary treatment is carried out to textile image, gaussian filtering, gray scale including mask size for 3*3
The gaussian filtering that processing and mask size are 7*7.
4. the fabric defects detection method according to claim 2 based on LBP and GLCM, which is characterized in that the step
(12) the GLCM feature extracting method based on LBP in extracts feature vector specifically:
(a) it usesImage gray levels are reduced to 0-P+1 by operator, whereinOperator is with rotational invariance
The local binary pattern operator of " unification " mode, P indicate pixel number in center pixel vertex neighborhood, and R indicates the radius of neighbourhood；
(b) gray level co-occurrence matrixes of 0 °, 45 °, 90 ° and 135 ° four direction are generated on this basis；
(c) average value of four gray level co-occurrence matrixes is calculated as mixing gray level co-occurrence matrixes；
(d) characteristic value of mixing gray level co-occurrence matrixes is extracted as feature vector.
5. the fabric defects detection method according to claim 4 based on LBP and GLCM, which is characterized in that the step
(d) feature vector in includes: energyEntropyContrastUnfavourable balance away fromAutocorrelationInconsistencyCluster shadeWith significant clusterWherein,P (i, j) indicates that gray value is respectively the pixel of i and j in gray scale symbiosis
Entry values in matrix, L indicate image gray levels.
6. the fabric defects detection method according to claim 1 based on LBP and GLCM, which is characterized in that the step
(2) include following sub-step:
(21) textile image to be detected is pre-processed, completes the filtering operation of image；
(22) detection window identical with template characteristic extraction stage size is chosen, according to fixed step-length m, first from left to right again
From downward overlapping sliding detection；
(23) feature vector of detection window is extracted using the GLCM feature extracting method based on LBP；
(24) feature vector and standard feature threshold value that will test window are compared, and carry out flaw window according to decision condition
Calibration.
7. the fabric defects detection method according to claim 6 based on LBP and GLCM, which is characterized in that the step
(21) pretreatment mode in is identical with the pretreatment mode in template characteristic extraction stage；Based on LBP's in the step (23)
GLCM feature extracting method is identical with the extracting method that template characteristic extracts the stage.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910074036.7A CN109858485A (en) | 2019-01-25 | 2019-01-25 | A kind of fabric defects detection method based on LBP and GLCM |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910074036.7A CN109858485A (en) | 2019-01-25 | 2019-01-25 | A kind of fabric defects detection method based on LBP and GLCM |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109858485A true CN109858485A (en) | 2019-06-07 |
Family
ID=66896175
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910074036.7A Pending CN109858485A (en) | 2019-01-25 | 2019-01-25 | A kind of fabric defects detection method based on LBP and GLCM |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109858485A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110349132A (en) * | 2019-06-25 | 2019-10-18 | 武汉纺织大学 | A kind of fabric defects detection method based on light-field camera extraction of depth information |
| CN114913512A (en) * | 2022-07-15 | 2022-08-16 | 南通浩盛汽车科技有限公司 | Automobile seat leather classification method, device and system based on computer vision |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102221559A (en) * | 2011-03-05 | 2011-10-19 | 河海大学常州校区 | Online automatic detection method of fabric defects based on machine vision and device thereof |
| CN103955922A (en) * | 2014-04-17 | 2014-07-30 | 西安工程大学 | Method for detecting flaws of printed fabric based on Gabor filter |
| CN104866862A (en) * | 2015-04-27 | 2015-08-26 | 中南大学 | Strip steel surface area type defect identification and classification method |
| CN105277567A (en) * | 2015-05-21 | 2016-01-27 | 南通大学 | Fabric flaw detection method |
| CN105654123A (en) * | 2015-12-26 | 2016-06-08 | 河北工业大学 | Identification method of woven fabric tissue chart |
| CN106097360A (en) * | 2016-06-17 | 2016-11-09 | 中南大学 | A kind of strip steel surface defect identification method and device |
| CN107274409A (en) * | 2017-06-21 | 2017-10-20 | 江阴芗菲服饰有限公司 | woven fabric defect segmentation method |
-
2019
- 2019-01-25 CN CN201910074036.7A patent/CN109858485A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102221559A (en) * | 2011-03-05 | 2011-10-19 | 河海大学常州校区 | Online automatic detection method of fabric defects based on machine vision and device thereof |
| CN103955922A (en) * | 2014-04-17 | 2014-07-30 | 西安工程大学 | Method for detecting flaws of printed fabric based on Gabor filter |
| CN104866862A (en) * | 2015-04-27 | 2015-08-26 | 中南大学 | Strip steel surface area type defect identification and classification method |
| CN105277567A (en) * | 2015-05-21 | 2016-01-27 | 南通大学 | Fabric flaw detection method |
| CN105654123A (en) * | 2015-12-26 | 2016-06-08 | 河北工业大学 | Identification method of woven fabric tissue chart |
| CN106097360A (en) * | 2016-06-17 | 2016-11-09 | 中南大学 | A kind of strip steel surface defect identification method and device |
| CN107274409A (en) * | 2017-06-21 | 2017-10-20 | 江阴芗菲服饰有限公司 | woven fabric defect segmentation method |
Non-Patent Citations (5)
| Title |
|---|
| ROBERT M.HARALICK: "Statistical and Structural Approaches to Texture", 《PROCEEDINGS OF THE IEEE》 * |
| 屈博: "布匹瑕疵检测技术研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
| 崔玲玲: "布匹瑕疵识别中的关键技术研究", 《中国博士学位论文全文数据库 信息科技辑》 * |
| 李玥灵 等: "基于LBP-GLCM纹理特征提取的服装图像检索", 《电视技术》 * |
| 李鑫 等: "基于机器视觉的智能化布匹瑕疵检测方法", 《河南工程学院学报（自然科学版）》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110349132A (en) * | 2019-06-25 | 2019-10-18 | 武汉纺织大学 | A kind of fabric defects detection method based on light-field camera extraction of depth information |
| CN110349132B (en) * | 2019-06-25 | 2021-06-08 | 武汉纺织大学 | Fabric flaw detection method based on light field camera depth information extraction |
| CN114913512A (en) * | 2022-07-15 | 2022-08-16 | 南通浩盛汽车科技有限公司 | Automobile seat leather classification method, device and system based on computer vision |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108961217B (en) | Surface defect detection method based on regular training | |
| CN114549522B (en) | Textile quality detection method based on target detection | |
| WO2020137222A1 (en) | Defect inspecting device, defect inspecting method, and program for same | |
| CN108765402B (en) | Non-woven fabric defect detection and classification method | |
| CN108022233A (en) | A kind of edge of work extracting method based on modified Canny operators | |
| CN109509171A (en) | A kind of Fabric Defects Inspection detection method based on GMM and image pyramid | |
| CN109118471A (en) | A kind of polishing workpiece, defect detection method suitable under complex environment | |
| CN109886960A (en) | The method of glass edge defects detection based on machine vision | |
| CN110781913B (en) | Zipper cloth belt defect detection method | |
| CN108647706A (en) | Article identification classification based on machine vision and flaw detection method | |
| CN109858485A (en) | A kind of fabric defects detection method based on LBP and GLCM | |
| CN111161237A (en) | Fruit and vegetable surface quality detection method, storage medium and sorting device thereof | |
| Kiruthika et al. | Matching of different rice grains using digital image processing | |
| CN105631481B (en) | Iron based on genetic programming composes abrasive grain compound characteristics building method | |
| CN111667475A (en) | Machine vision-based Chinese date grading detection method | |
| CN115205209A (en) | Monochrome cloth flaw detection method based on weak supervised learning | |
| CN102980659A (en) | Digitalized characterization method of monochrome tight fabric surface color | |
| CN112435235A (en) | Seed cotton impurity content detection method based on image analysis | |
| Gurubelli et al. | Texture and colour gradient features for grade analysis of pomegranate and mango fruits using kernel-SVM classifiers | |
| CN114998290A (en) | Fabric flaw detection method, device, equipment and medium based on supervised mode | |
| CN110728286A (en) | Abrasive belt grinding material removal rate identification method based on spark image | |
| CN107862689A (en) | Leather surface substantially damaged automatic identifying method and computer-readable recording medium | |
| CN110570418A (en) | Woven label defect detection method and device | |
| Bodnarova et al. | A constrained minimisation approach to optimise Gabor filters for detecting flaws in woven textiles | |
| Jothi et al. | Intra-ocular lens defect detection using generalized hough 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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190607 |