Professional Documents
Culture Documents
Abstract
Fabric defect detection is a necessary and essential step of quality control in the textile
manufacturing industry. This paper has been reviewed the various fabric defect detection and
classification methods of statistical, spectral, model based and structural approaches. This paper has been
presented the survey on types of defects, detection accuracy, performance metric and inference from
recent publications. It will benefit researchers and practitioners in image processing and computer vision
fields in understanding the characteristics of the different defect detection approaches. It concludes that
the pulse coupled neural network (PCNN) approach is better detection accuracy than the other methods
and is suggested for further research.
[14], [17], [20], [21], [25], [29], [30] and [34] in model. Particularly, model-based approaches are
Table 1. suitable for fabric inspection when the statistical and
spectral approaches have not yet shown their utility. The
2.2. Spectral approaches advantage of the modeling is that it can produce textures
that can match the observed textures. Model based
The primary objectives of these approaches are approaches are particularly suitable for fabric images
firstly to extract texture primitives, and secondly to with stochastic surface variations and this method was
model or generalize the spatial placement rules. Spectral reviewed in serial no of [21], [27] and [32] in Table 1.
approaches require a high degree of periodicity thus, it
is recommended to be applied only for computer vision 2.4. Structural approaches
of uniform textured materials like fabrics. Spectral
approach enabled to capture the most outstanding In structural approaches, texture is
features of all defects and provide multi resolution of characterized by texture primitives or texture elements,
image. This method was used for image quality and the spatial arrangement of these primitives. These
enhancement, noise reduction and feature extraction. primitives can be as simple as individual pixels, a region
For automated defect detection, Spectral approaches with uniform gray levels, or line segments.
occupy a big part of the latest computer vision research Consequently, the main objects of these approaches are
work and this method was reviewed in serial no of [7], firstly to extract texture primitives, and secondly to
[19], [23], [24], [26] and [35] in Table 1. model or generalize the spatial placement rules. The
Placement rules can be obtained through modeling
2.3. Model based approaches geometric relationships between primitives or learning
statistical properties from texture primitives. However,
In this approach, the texture is regarded as a these approaches were not successful on fabric defect
complex pictorial pattern and can be defined by a detection, mainly due to the stochastic variations in the
deterministic model. Model-based texture analysis fabric structure (due to elasticity of yarns, fabric motion,
methods try to generate the texture. Here model the fiber heap, noise, etc.) and this method was reviewed in
texture by determining the parameters of a pre-defined serial no of [22] and [23] in Table 1.
Bi-level Thresholding Fourier transform Gauss markov random field Primitive measurement
Gray level statistics Gabor filtering Poisons model Edge features
Morphological operations Wavelet transform Model based clustering
Auto correlation Weiner distributions
Fractal method Windowed fourier transform
Normalized cross correlation Optimized FIR filter
Statistical moments
Multi level thresholding
rank order functionz
21
M. Fathu Nisha et al. /J. Environ. Nanotechnol., Vol. 6(2), 20-29, (2017)
Type of
Detection and Input Detection Performance
S.No defects Inference
classification technique image accuracy metric
detected
Kernel principal Uniform Hole, oil spot, Detection This method achieved
component analysis & thread error, success rate, high true detection rate
3. 96%
(KPCA) (Junfeng Jing et Structured objects on the specificity, and low cost for online
al. 2016) fabric surfaces sensitivity fabric inspection.
Form factor,
rectangularity This method not
Feature extraction factor, location, required any
Color Needle
technique, graph based orientation, adjustment, finding the
6. fabric cutting, oil 89%
segmentation (Prasad intensity based better accuracy and less
image stain
dakhole et al. 2016) features, time consumption in the
moment based industry.
features
Yarn distance,
This method determined
energy
yarn paths even in
distribution
2D Fast fourier highly draped border
Hole, parameter,
transforms (FFT) Woven regions of the surface.
7. excessive 78% maximum
(George bardi et al. fabric So the influence of
margin energy
2016) different settings on
parameter,
fiber orientation is
displacement,
easily evaluated.
speed
22
M. Fathu Nisha et al. /J. Environ. Nanotechnol., Vol. 6(2), 20-29, (2017)
Mispick, Reconstruction
Broken pick, error, sum of the
buckling, tom absolute value
The algorithm was run
selvedge, of projection,
with high parallel
rough largest absolute
Small scale over efficiency. detection
selvedge, value of
completed dictionary of Twill result was easily
8. fuzzy ball, 93.7% projection,
sparse coding (SSOCD) fabric affected by noise
bore, yarn, distance
(Feng et al. 2016) through, illumination
double flat, between the
changing, fabric grain
yarn evenness, image, average
fluctuating, etc.
pulled in of trained
selvedge, oil samples, Size of
warp image patches
Patterned
fabric
Oil stain, stab
image – Image decomposition
cut, hole, Divergence
Artificial neural network star with ANN classifier
broken end, factor, mean
11. (ANN) (karunamoorthy pattern, 95% provided best defect
netting standard
et al. 2015) dot detection result among
multiple, thin deviation
pattern, all testing methods
bar,
box
pattern
Inter pixel
distance, inter
Holes, weft
pixel
crackiness, High accuracy, better
Auto correlation function Yarn dyed orientation,
14. broken weft, 90% adaptability is provided
(Dandan zhu et al. 2015) fabric euclidean
wrong weft, by this method
distance, size of
oil stain
detection
window,
23
M. Fathu Nisha et al. /J. Environ. Nanotechnol., Vol. 6(2), 20-29, (2017)
Flaw area,
dispersion, This method took 0.2
aspect ratio, seconds to detect the
Oil warp, oil euler number, flawed areas and
Rough set classifier Regular
weft, missing maximum of classify the type of each
15. (Sudarshan Deshmukh et patterned 96%
pick, thin bar, gray value, flawed area. This
al. 2015) fabric
thick bar minimum of system suitable for on
gray value, line real time
mean of gray application.
value,
Mean, energy,
standard
Artificial neural network Plain High detection accuracy
deviation,
20. (ANN) (Banumathi et al. woven Hole , oil stain 89% is provided by this
entropy,
2015) fabric method.
skewness,
kurtosis,
24
M. Fathu Nisha et al. /J. Environ. Nanotechnol., Vol. 6(2), 20-29, (2017)
Area of
Vertical defective Fast detection and
Canny edge detection missing yarn, region, total extraction of fabric
Textile
22. (Halimi Abdellah et al. horizontal 94.84% area of window defects from the images
fabric
2014) missing yarn, size, , centroid of textile fabric is done
hole, spot of defective by this method
parts
Average gray
Warp lacking, 83% for The defective areas of
Un dyed level, average
Discrete wavelet weft lacking, off line the fabric sample are
denim contrast,
23. transform DWT)(Ibrahim soiled yarn, detected successfully
fabric 88% for smoothness,
Celik et al. 2014) hole, knot or and the boundaries of
sample real time entropy,
yarn flow them are labeled.
correlation
Energy,
No defect,
contrast,
weaving defect
correlation This method is simple
in weft
Discrete cosine transform Plain & 97% cluster to understand. This
direction,
24. (DCT) (Semiyaa et al. Patterned prominence, method was give good
weaving defect
2014) fabric horizontal , defect detection
in warp
vertical , performance.
direction,
diagonal energy
holes, oil spot
value
proposed method is a
Broken end, robust and accurate
Bi dimensional empirical
Synthetic broken pick, Intrinsic mode
28. mode decomposition 96.1% approach for fabric
fabric bars, stitches, function fusion
(Zhengxin li, et al. 2014) defect inspection.
foreign fiber,
Gabor
Uniform, Hole, oil spot,
Principal component generating This method is reduced
structured thread error,
29. analysis (PCA) (Lucia 98.1% function, kernel computational cost up
fabric, objects on the
Bissi et al. 2013) size, correction to 33% e.
wool surfaces
gain
25
M. Fathu Nisha et al. /J. Environ. Nanotechnol., Vol. 6(2), 20-29, (2017)
In this method,
Stain, hole,
Feed forward neural Plain inspection time was
weft float type Mean, standard
30. network (FFNN) (Nasira woven 94.6% reduced, high accuracy
fault, warp deviation
et al. 2013) fabric than manual inspection
float type fault
system
Ali Rebhi, SabeurAbid and FarhatFraeich, Texture Georg Bardi, Andreas Nocke, Chokri Cherif, Matthias
defect detection using local homogeneity and Pooch, Martin Schulze, Henning Heuer, Marko
discrete cosine transform, World Applied sciences Schiller, Richard Kupke and Marcus Klein,
Journal, 31(9), 1677-1683(2014). Automated detection of yarn orientation in 3D
doi:10.5829/idosi.wasj.2014.31.09.156 draped carbon fiber fabrics and performs from eddy
current data, Composites Part B: Egineering, 96,
Alam, F. I. and Faruqui, U., Optimized Calculations of
312-324(2016).
Haralick Texture Features, European, J., of doi:10.1016/j.compositesb.2016.04.040
Scientific Research, 543-553(2011). Gangandeep Singh, Gurpadam Singh and Mandeep
Asha, V., Bhajantri, N. U. and Nagabhushan, P., GLCM Kaur, Fabric Defect Detection using series of image
based chi-square histogram distance for automatic Processing Algorithm & ANN operation, Global J.,
detection of defects on patterned textures, Int. J. Comp. Tech., 4(2), 225-228(2016).
Computational Vision and Robotics, 2(4), 302- Gede Surya Rahayuda, Texture Analysis on image
313(2011). Motif of Endek bali using K-nearest neighbor
doi:10.1504/IJCVR.2011.045267 Classification method, Int. J. Adv. Comp. Sci.
Ajay Kumar, Member, Computer vision based fabric Appl., 6(9), 205-211(2015).
defect detection: A survey, IEEE Trans. Industrial Halimi Abdellah, Roukhe Ahmed and Ouhmad
Elec., 55(1), 348-363(2008). Slimane, Defect detection and identification in
doi: 10.1109/tie.1930.896476 textile fabric by SVM Method, IOSR J. Eng., 4(12),
69-77(2014).
Bhavini Patel and Hetal Bhaidasna, Survey on different
methods for Defect Detection, Int., Res. J. Engg. Halimi Abdellah, El kouraychi abdelmajid, Bouzzid
Abdenabi and Roukhe Ahmed, Defects detection
Tech., 3(2), 1217-1220(2016)
and extraction in textile imageries using
Basavaraj S. Anami and Mahantesh C. Elemmi, A rule
mathematical morphology and geometrical features,
based approach for classification of shades of basic
J., Signal Processing Theory and Applications, 1,
colors of fabric images, Int. J. of Signal Processing,
01-16(2012).
Image Processing and Pattern Recognition, 8(2),
doi:10.7726/jspta.2012.1001
389-400(2015).
Ibrahim Celik, Cannan Dulger and Mehmet
doi:10.14257/ijsip.2015.8.2.36 Topalbekiroglu, L., Fabric defect detection using
Basturk, K., Ketencioglu, H. and Yugnak, Z., linear filtering and morphological operations, Indian
Inspection of defects in fabrics using Gabor wavelets J. Fibre and Textile Research, 39(3), 254-
and principle component analysis, Signal Processing 259(2014).
and its Applications, 01-04(2007). Junfeng Jing, Xiaoting Fan and Penegfei Li, Automated
doi:10.1109/ISSPA.2007.4555353 Fabric Defect Detection based on Multiple Gabor
Choudhary, D., Ajay Kumar Singh, Shamik Tiwari and Filters and KPCA, Int. J. of Multimedia and
Shukla, V. P., Performance Analysis of Texture Ubiquitous Eng., 11(6), 93-106, (2016).
Image Classification Using Wavelet Feature, Int., J., doi:10.14257/ijmue.2016.11.6.09
Image, Graphics and Signal Processing, 1, 58- Jagdish Lal Raheja, Bandla Ajay and Ankit
63(2013). Chaudhary, Real time fabric defect detection system
doi:10.5815/ijigsp.2013.01.08 on an embedded DSP platform, Int. J. Light and
Dandan Zhu, Ruru Pan and Wiedong Gao, Yarn dyed Electron Optics, Elsevier(2013).
fabric defect detection based on autocorrelation doi: 10.1016/j.ijleo.2013.03.038
function and GLCM, AUTEX Research J., 15(3), Jagdish Lal Raheja, Sunil Kumar, Fabric defect
226-232(2015). detection based on GLCM and Gabor filter: A
doi:10.1515/aut-2015-0001 comparison, 124(23), 6469– 6474(2013).
Eldessouki, M., Hassan, M., Qashqari, K. and Shady, doi:10.1016/j.ijleo.2013.05.004
E., Application of principal component analysis to Jayanta K. Chandra, Pradipta K. Banerjee and Asit K.
boost the performance of an automated fabric fault Datta, Neural network trained morphological
detector and classifier, Fibers and Textiles in processing for the detection of defects in woven
Eastern Europe, 51-57(2014). fabric, J. Textile Institute, 101(8), 699-706(2010).
Farsi, H. and Mohamadzadeh, S., Color and texture doi: 10.1080/00405000902812735
feature based Image retrieval by using hadamard Karunamoorhy, B., Somandareswari, D. and Sethu, S.
matrix in discrete wavelet transform, Image P., Automated Patterned Fabric Fault Detection
Processing, 7(3), 212-218(2013). using Image Processing Technique in Matlab, Int. J.
doi:10.1049/iet-ipr.2012.0203 Adv. Res.Comp. Eng. Tech., 4(1), 63-69(2015).
27
M. Fathu Nisha et al. /J. Environ. Nanotechnol., Vol. 6(2), 20-29, (2017)
Le Tong, Wong, W. K. and Kwong, C. K., Differential Revathy, G. and Vidhyalakshmi, P., Defect
evolution based optimal Gabor filter model for classification in fabric web material using lab view,
fabric inspection, J. Neurocomputing, 173(3), 1386- Int. J. Comp. Appl., 63(18), 44-48( 2013).
1401(2016). Sudarshan Deshmukh, S., Raut, S. S. and Biradar, M.
doi: 10.1016/j.neucom.2015.09.011 S., Defect detection of regular patterned fabric by
Lei Zhang, Junfeng Jing and Hongwei Zhang, Fabric spectral estimation technique and rough set
defect classification based on LBP and GLCM, J., classifier, Int. J. Innovative Sci. Engg. Tech., 2(4),
Fiber Bioengineering and Informatics., 8(1), 81- 760-768(2015).
89(2015). Semiyya, E., Jothi, R. L. and Najuma, E., Automatic
doi:10.3993/jfbi03201508 fabric defect detection using spatial and spectral
Md. Ashraful Alam, Naushad Ali, M. M., Musaddeque features, Int. J. Engg. Res. Tech., 3(12), 1019-
Anwar Al-Abedin Syed, Nawaj Sorif and Md. Abdur 1023(2014).
Rahaman, An algorithm to detect and identify Schneider, and Dorit Merhof, D., Blind Weave
defects of industrial pipes using image processing, Detection for Woven fabrics, Springer, Dorian
IEEE.(2014). Pattern Anal. Appl., 725-737(2014).
doi:10.1109/SKIMA.2014.7083567 Sheetal Thorave and Biradar, M. S., New technology
Marini, A., Facon, J. and Koerich, A. L., Bird species for fabric defect detection based on K-means
classification based on color features, IEEE algorithm, IOSR J. Engg., 4(3), 15-20(2014).
International Conference, 4336-4341, (2013). Szczypiński, P., Klepaczko, A., Pazurek, M. and
doi:10.1109/smc.2013.740 Daniel, P., Texture and color based Image
Marcos Yutaka Shiinoa, Maria Odila Hilário Cioffia, Segmentation and Pathology detection in Capsule
and Herman Jacobus Cornelis Voorwald,
endoscopy videos, Computer methods and programs
Evaluation of stitched fabric composite processed by
in biomedicine, 113(1), 396-411(2014).
RTM in Quasi-static test, Science Direct Procedia
doi:10.1016/j.cmpb.2012.09.004
Engineering, 10, 2603–2608(2011).
Sabeenian, R. S., Paramasivam, M. E. and Dinesh, P.
doi:10.1016/j.proeng.2011.04.434
Meihong Shi, Rong Fu, Yong Guo, Fabric defect M., Computer vision based defect detection and
detection using local contrast deviations, Springer, identification in handloom silk fabrics, Int. J. of
52, 147-157(2011). Comp. Appl., 42(17), 41-48(2012).
Nasira and Banumathi, P. M., Plain woven fabric defect Swati F. bhope and Patil, S. P., Defects Identification in
detection based on image processing and artificial Textile Industries, Int., J., of Chemical Sci.,and
neural networks, Int. J. Comp. Trends Tech., 6(4), Appl., 37-41(2010).
226-229(2013). Tiangpeng Feng, Lian Zou, Jia Yan, Wenxuan Shi,
Ngan, H. Y. T., Pang, G. K. H. and Yung, N. C. H., Yifeng Liu, Cien Fan and Dexiang Deng, Real time
Automatic Fabric Defect Detection a Review, Image fabric defect detection using accelerated small scale
and Vision Computing, 29, 442-458(2011). over completed dictionary of sparse coding, Int. J.
doi:10.1016/j.imavis.2011.02.002 Advanced Robotic Systems., 13(1), 01-09(2016).
Prasad Dakhole, Shruti kalode and Ankit Patel, Fabric doi: 10.5772/62058
fault detection using image processing matlab, Int. Vikrant Tiwari and Gaurav Sharma, Automatic fabric
J. Emerging Trends Engg. Management Res., 2(1), fault detection using morphological operations on bit
(2016). plane, Int. J. Comp. Sci. Network Security, 15(10),
Pritpal Singh and Prabhjyot Singh, Texture analysis in 30-35(2015).
Fabric material for Quality evaluation using GLCM Vaibhav V. Karlekar, Biradar, M. S. and Bhangale, K.
matrix, Int. J. Applied Engg. Tech., 5(1), 01- B., Fabric Defect Detection using Wavelet Filter,
05(2015). International Conference on Computing
Parmeetkaur Randhawa, Uppal and Raju Sharma, Communication Control and Automation, IEEE.,
Automatic detection and classification of fabric (2015).
defects using large margin nearest neighbor method, doi:10.1109/ICCUBEA.2015.145
Int. J. Tech. Engg. Sci., 2(6), 1943-1946(2014). Xianghua Xie, Review of recent advances in surface
Priyanka M. Shanbhag, Manish P. Deshmukh and defect detection using texture analysis techniques,
Shekhar R. Suralkar, Overview: Methods of Electronic Letters on Computer Vision and Image
Automatic Fabric Defect Detection, Global J. Eng. Analysis, 7(3), 01-22(2008).
Design Technol., 1(2), 42-46(2012). Yundong Li, Weigang Zhao and Jiahao Pan,
Pan R. R., Weidong, QIAN Xinxin, ZHANG Xiaoting, Deformable Patterned Fabric Defect Detection with
Automatic analysis of woven fabric weave based on Fisher Criterion based Deep Learning, IEEE Trans.
notation, J. Textile Res., 30(11), 30-34(2010). on Automation Sci. Engg., 01-09(2016).
28
M. Fathu Nisha et al. /J. Environ. Nanotechnol., Vol. 6(2), 20-29, (2017)
Yin KaiCheng. and Yu Weidong, Clothing Defect Yousif Mohamed Y. Abdulla, Ali Hayder and Eltayeb
Detection carbon-fibre-reinforced plastics (CFRP) Wagiallah, Automatic Enhancement of
using Multi-frequency Eddy current sensors, IEEE Mammography Images using Contrast Algorithm,
Instrum. Meas., Technol. Conf. Proc.,(2007). Int. J. Sci. Res., 3(9), 1885-1889(2012).
29