See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/349161174

Fabric Defect Detection System

Conference Paper in Advances in Intelligent Systems and Computing · February 2021

DOI: 10.1007/978-3-030-68154-8_68

CITATIONS READS

33 6,131

4 authors, including:

Tanjim Mahmud Juel Sikder

Rangamati Science and Technology University Rangamati Science and Technology University, Rangamati
131 PUBLICATIONS 2,112 CITATIONS 32 PUBLICATIONS 360 CITATIONS

SEE PROFILE SEE PROFILE

Rana Joyti Chakma

Rangamati Science and Technology University
16 PUBLICATIONS 307 CITATIONS

SEE PROFILE

All content following this page was uploaded by Tanjim Mahmud on 15 September 2023.

The user has requested enhancement of the downloaded file.

 Fabric Defect Detection System
Author Proof

Tanjim Mahmud1(&), Juel Sikder1, Rana Jyoti Chakma1,

and Jannat Fardoush2
1
Department of Computer Science and Engineering,
Rangamati Science and Technology University, Rangamati, Bangladesh
{tanjim.cse,rchakma}@rmstu.edu.bd,
[email protected]
2
Department of Computer Science and Engineering, University of Chittagong,
Chittagong, Bangladesh
[email protected]

Abstract. Fabric inspection is very significant in textile manufacturing. Quality AQ1

of fabric defends on vital activities of fabric inspection to detect the defects of
fabric. Profits of industrialists have been decreased due to fabric defects and
cause disagreeable loses. Traditional defect detection methods are conducted in
many industries by professional human inspectors who manually draw defect
patterns. However, such detection methods have some shortcomings such as
exhaustion, tediousness, negligence, inaccuracy, complication as well as time-
consuming which cause to reduce the finding of faults. In order to solve these
issues, a framework based on image processing has been implemented to
automatically and efficiently detect and identify fabric defects. In three steps, the
proposed system works. In the first step, image segmentation has been employed
on more than a few fabric images in order to enhance the fabric images and to
find the valuable information and eliminate the unusable information of the
image by using edge detection techniques. After the first step of the paper,
morphological operations have been employed on the fabric image. In the third
step, feature extraction has been done through FAST (Features from Accelerated
Segment Test) extractor. After feature extraction, If PCA (Principal Component
Analysis) is applied as it reduces the dimensions and preserves the useful
information and classifies the various fabric defects through a neural network
and used to find the classification accuracy. The proposed system provides high
accuracy as compared to the other system. The investigation has been done in a
MATLAB environment on real images of the TILDA database. AQ2

Keywords: Defect detection
FAST (Features from Accelerated Segment

Test)
Neural network
PCA (Principal Component Analysis)

1 Introduction

The textile industry is a rising sector. Development and advancement of the sector
normally bring to build the going through huge investment. Be that as it may, the
textile, like any other sector, industry experienced various issues. These include some
insurance to diminish the effect of misfortunes that are budgetary, client disappoint-
ment, time squandering, and so on. Fabric defects are probably the greatest test
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2021
P. Vasant et al. (Eds.): ICO 2019, AISC 1324, pp. 1–13, 2021.
https://doi.org/10.1007/978-3-030-68154-8_68
 2 T. Mahmud et al.

confronting the textile business. Fabric is made in a day by day life utilizing fibers and
Author Proof

a usually utilized material. Most fabrics are delivered after passing through a series of
making stages. Various machines and methods are utilized during the making stages.
Fabrics are subjected to pressures and stresses along these lines that cause defects. As
indicated by their structures and directions, defects take various names. The textile
business has distinguished in more than 70 types of defects [1] such as laddering, end-
out, hole, and oil spot as shown in Fig. 1. Unexpected tasks might be the reason for
various defects on the fabric surface during the manufacturing of fabric [2]. The lack of
defects can diminish the cost of fabric by 50–60% [1]. The decrease in the impacts in
the production process is typical for the industrialist.

Fig. 1. Different defects in a fabric

Thus, fabric manufacturing is one of the largest traditional businesses where fabric
inspection systems can play a vital role in growing the manufacturing rate. These days,
the significance of an inspection process nearly rises to with manufacturing process in
the current industrialist viewpoint. The idea of inspection process is to recognize the
happened errors or defects, on the off chance that any exist, at that point to change
argument or give alert of inspector for checking the manufacturing procedure [3]. For
the most part, fabric defects recognition utilizes two kinds of investigation models [4].
The essential one is that the human-based inspection systems as shown in Fig. 2.The
second framework is automated based inspection systems as shown in Fig. 3.
Accordingly, human-based defect detection done by specialists’ turns out to be rapidly
a mind-boggling and fussy task [5, 6]. In this manner, having proficient and automated
based frameworks nearby is a significant necessity for improving unwavering quality
and accelerating quality control, which may expand the profitability [7–10]. The
subject of automated based defect detection has been examined in a few works in the
most recent decades. In spite of the fact that there is no widespread methodology for
handling this issue, a few strategies dependent on image processing procedures have
been proposed in recent years [11–13]. These strategies were utilized to recognize
defects at the image level, so the precision rate is little and additionally, it is hard to find
 Fabric Defect Detection System 3

the defects precisely. In this way, they can’t be stretched out to various fabrics. As of
Author Proof

late, some different techniques dependent on local image-level have been proposed,
which utilize the base unit as the fundamental activity object to extract image features.
These methodologies can be ordered into four principle gatherings: Statistical,
Signal processing-based, Structural methodology, and Model-based methodology.

Fig. 2. Human-based inspection system Fig. 3. Machine automated inspection system

In the statistical approach, gray-level properties are utilized to describe the textural
property of texture image or a measure of gray-level reliance, which are called 1st-order
statistics and higher-order statistics, separately [14]. The 1st-order statistics, for
example, mean and standard deviation [15, 16], rank function [17], local integration,
can gauge the variance of gray-level intensity among different features between
defective areas and background. The higher-order statistics depends on the joint
probability distribution of pixel sets, for example, gray-level co-occurrence matrix [18]
gray-level difference strategy [15] and autocorrelation method. In any case, the
inconvenience of this strategy is that defects size is sufficiently enormous to empower a
compelling estimation of the texture property. So this methodology is feeble in han-
dling local little defects. Additionally, the calculation of higher-order statistics is
tedious [17].
In the subsequent class model-based methodology, the generally utilized strategies
are Markov random field Gaussian Markov random field [16]. The texture features of a
contemplated texture and can signify to all the more exactly spatial interrelationships
between the gray-levels in the texture. However, like the methodologies based on
second-order statistics, additionally it is tough for model-based methodology to deal
with identifying small-sized defects in light of the fact that the methodologies as a rule
require an adequately large region of the texture to assess the parameters of the models.
The structural approach generally utilized on properties of the primitives of the
defect-free fabric texture for the nearness of the flawed region, and their related
placement rules. Apparently, the practicability of this methodology is to congestion to
those textures with regular macro texture.
 4 T. Mahmud et al.

Not at all like the above methodologies which separate the defects as far as the
Author Proof

visual properties of the fabric texture, the signal processing based methodology extract
features by applying different signal processing procedures on the fabric image. It is
projected that the distinguishable between the defect and the non-defect can be
improved in the handled fabric image. This methodology further comprises of the
accompanying techniques: Spatial filtering, Karhunen-Loeve transform, Fourier
transform, Gabor transform, and Wavelets transform.
As a weakness of this methodology, its performance is effortlessly influenced by
the noise in the fabric image. These coefficients exemplify to optimal the defect-free
fabric image, be that as it may, not the optimal separation between the defect and the
non-defect.
They are progressively proficient in the separation of fabric defects than different
techniques that depend on the texture investigation at a single scale [19]. Contrasted
with the Gabor transform; the wavelet transform has the benefit of greater adaptability
in the decomposition of the fabric image [20]. Consequently, the wavelet transform is
seen as the most suitable way to deal with the feature extraction for fabric defect
detection.

Table 1. Taxonomy of some most recent related works

Article Classifier Machine learning technique Accuracy rate
[21] Artificial Neural Network Counterpropagation 82.97%
[22] Artificial Neural Network Backpropagation 78.4%
[23] Artificial Neural Network Resilient backpropagation 85.57%
[24] Support Vector Machine NA 77%
[25] Artificial Neural Network Backpropagation 84%
[26] Artificial Neural Network Backpropagation 81%
Artificial Neural Network Least mean square error (LMS) 87.21%
[27] Artificial Neural Network Backpropagation 85.9%
[28] Artificial Neural Network Backpropagation 76.5%
[29] Artificial Neural Network Learning vector quantization (LVQ) 67.11%
[30] Model-based clustering NA 65.2%
[31] Artificial Neural Network Backpropagation 71.34%
[32] Artificial Neural Network Resilient backpropagation 69.1%

Table 1 Illustrates the taxonomy of most recent fabric defects detection methods, in
light of their classifier, machine learning technique, and accuracy rate.
In this paper, we propose an innovative defect detection algorithm which has the
capability to cope with different types of defects. Our algorithm is based on four
phases. In the initial phase, image segmentation has been utilized on an excess of a
couple of fabric images so as to enhance the fabric image and to locate the important
data and wipe out the unusable data of the image by utilizing different edge detection
strategies. After the initial phase, morphological operations have been utilized on the
 Fabric Defect Detection System 5

fabric image. In the third step, feature extraction has been done through FAST (Fea-
Author Proof

tures from Accelerated Segment Test) extractor. After feature extraction, If PCA is
applied as it lessens the dimensions and preserves the helpful data, and characterizes
the different fabric defects through a neural network; additionally classifier has been
utilized to find the accuracy rate. The proposed framework gives high precision when
contrasted with the other framework. The investigation has been done in MATLAB
environment on real images of the TILDA database [33].
The remaining of the paper is arranged as follows: In Sect. 2, the various types of
fabric defects are presented. In Sect. 3 explains our proposed approach for defect
detection. In Sect. 4 presents the application of our system and analysis. Finally, Sect. 5
accomplishes the paper and presents our future research plans.

2 Defects in Fabric

In order to prepare various categories and forms of fabric items in the industry, fabric
materials are used. Consequently, yarn quality and/or loom defects affect the fabric
quality. Fabric defect has been estimated [34] that the price of fabrics is reduced by
45%-65% due to the presence of defects such as dye mark/dye Spot, slack warp, faulty
pattern card, holes, spirality, grease oil/ dirty stains, mispick, slub, wrong end, slack
end, and so on [1]. In a fabric, defects can occur due to: machine faults, color bleeding,
yarn problems, excessive stretching, hole, dirt spot, scratch, poor finishing, crack point,
material defects, processing defects, and so on [35, 36].

3 Proposed Methodology for Defect Detection

Fig. 4. Block diagram of the developed system

Figure 4 shows the steps of methodology, to sum up; the following steps are image
segmentation, feature extraction, PCA (Principal Component Analysis), and image
classification.
 6 T. Mahmud et al.

3.1 Image Segmentation

Author Proof

Image segmentation is a fundamental advance in image analysis. Segmentation isolates

an image into its objects or segment parts. Edge detection is a mechanism in image
processing to make the image segmentation procedure and pattern recognition more
precise [37, 38]. It fundamentally diminishes the measure of information and filters out
pointless data, while protecting the helpful properties in an image. The adequacy of
many image processing relies upon the flawlessness of identifying significant edges. It
is one of the procedures for detecting strength discontinuities in a digital image.
Essentially we can say, the way toward arranging and setting sharp discontinuities in an
image is known as edge detection. There are many edge detection strategies available,
every procedure intended to be keen on particular types of edges. Fac-
tors that are concerned about the choice of an operator for edge detection include
edge direction, edge structure, and noise condition. The paper applied the histogram
equalization strategy on fabric image as shown in Fig. 6. After that edge detection
strategy has been applied as shown in Fig. 6. There are numerous operators in edge
detection methods model Roberts, Sobel, and Prewitt [39, 40] and the result shows that
the Canny’s edge detection method performs superior to every other strategy.

3.2 Feature Extraction

The feature extractor applied on the dataset of images as shown in Fig. 6 which re-
lies on the extractor of the local feature. The point of local feature portrayal is to
express to the image which depends on some notable regions. The image relies upon its
local structures with a lot of local feature extractors and which is get from a lot of
image regions called interest regions [41]. FAST (Features from Accelerated Segment
Test) has been applied to the fabric picture to extract the features as shown in Fig. 6.
Rosten and Drummond proposed initially that FAST is a strategy for recognizing
interest regions in an image [42]. An interest region in an image is a pixel that has a
very much characterized position and can be vigorously identified. An interest region
has high local data substance and they ought to be in a perfect world repeatable
between various images [43].

3.3 PCA (Principal Component Analysis)

It is a straight forward method used in dimensionality reduction to decrease the features
that are not helpful. It protects the valuable features of the information or image [44, 45].
If applied after feature extraction through feature extractor FAST, it gives a great per-
formance or accuracy via classifier.
 Fabric Defect Detection System 7

3.4 Image Classification

Author Proof

Neural network [26, 45] is a machine learning technique that has been applied to
classify the Images of fabric defects as shown in Fig. 6 and applied the pattern
recognition framework and to get the great outcomes after train the framework by
dataset [46, 47]. It can partition the dataset into a testing stage and training stage to
locate the hidden neurons in the pattern recognition framework as shown in Fig. 5. The
classifier will apply for the classification accuracy when feature extraction from the
extractor. The classifier has been applied to the real image of fabric [33].

Fig. 5. Neural network

4 Application of the System and Analysis

4.1 Proposed Methodology
Phase For enhancing the fabric images to apply edge detection method
1
Phase For the application of FAST (Features from Accelerated Segment Test) extract the
2 features and discover the interest regions
Phase Consequently, applied the PCA (Principal Component Analysis) to decrease the
3 dimensions and reserve the beneficial data
Phase Applied machine learning algorithms neural network to get better accuracy as it
4 provides better outcomes
 8 T. Mahmud et al.
Author Proof

Original image Gray image Histogram Binary image Detected Region

Fig. 6. Application of the system

Fig. 7. MATLAB environment for defect detection

 Fabric Defect Detection System 9

Table 2. Comparison of neural network-based classification models

Author Proof

Article Accuracy Comment

[21] 82.97%
[22] 78.4%
[23] 85.57%
[24] 77%
[25] 84%
[26] 85.9%
[27] 76.5%
[28] 67.11%
[29] 65.2%
[30] 71.34%
[31] 69.1%
Developed system 97.21% Greatest accuracy among all developed systems

As compared to the other techniques our methodology provides better accuracy as

shown in Table 2.
Experiments are worked on TILDA database [33] and give better result in terms of
100% detection rate for the training set and 97.21% accuracy for the test set.

Fig. 8. System snapshot

 10 T. Mahmud et al.
Author Proof

Accuracy
120.00%
100.00%
80.00%
60.00%
40.00%
20.00% Accuracy
0.00%

Article

Fig. 9. Performance comparison of different studies

The proposed methodology feature extraction technique as compared to the other

discussed methodology feature extraction techniques using machine learning algo-
rithms gives better accuracy as shown in Fig. 9.

5 Conclusion and Future Work

The detection of faulty fabrics plays an important role in the success of any fabric
industry. The fabric industry needs a real-time quality control to find defects quickly
and efficiently. Manual control is inefficient and time-consuming that leads to heavy
loss. On the other hand, automatic quality control is considerably more proficient, in
light of the fact that it is a real-time and autonomous compared to manual productivity.
Till now all the fabric detection systems suggested by all the researchers, the accuracy
rate for detecting defective fabric is very low. However, this paper analyzed the
shortcomings of the traditional approach for fabric defect detection, and proposed an
innovative fabric defect detection technique based on a FAST (Features from Accel-
erated Segment Test) extractors and PCA (Principal Component Analysis) combined
with a neural network classification, to enhance the recognition accuracy rate that
texture fabrics cannot be effectively detected by present existing techniques. The paper
concludes that the proposed fabric defect detection technique gave better accuracy after
applied the machine learning algorithm and PCA in comparison to the other referenced
approaches. Additionally, our method notably showed its efficiency in separating
defect-free from defective areas of the fabric. Moreover, after a series of improvements,
our method exhibited better recognition performance for the fabric images. Having
successfully trained the neural network, 30 samples of each type of defect were used to
 Fabric Defect Detection System 11

assess the accuracy of the network classification. The defective images were then
Author Proof

graded with a 97.21% overall accuracy score. With a 100% accuracy score, the dye
spot defect was identified. The experimentation has been applied to the real images of
the TILDA database dataset. The implementation has been done in MATLAB software
as shown in Fig. 7 and Fig. 8. It automatically detects the fabric defect.
In the future, we will focus on sensor data-oriented systems and developing an
optimal system to match more accurately with a real system for fabric defect detection
as well as applying our machine learning algorithms for other different feature
extractors.

References
1. Stojanovic, R., Mitropulos, P., Koulamas, C., Karayiannis, Y., Koubias, S., Papadopoulos,
G.: Real-time vision-based system for textile fabric inspection. Real-Time Imaging 7, 507–
518 (2001)
2. Aasim A.: A catalogue of visual textile defects, ministry of textiles (2004)
3. Newman, T.S., Jain, A.K.: A survey of automated visual inspection. Comput. Vis. Image
Underst. 61(2), 231–262 (1995)
4. Kumar, A.: Computer-vision-based fabric defect detection: a survey. IEEE Trans. Ind.
Electron. 55(1), 348–363 (2008)
5. Huart, J., Postaire, J.G.: Integration of computer vision on to weavers for quality control in
the textile industry. In: Proceeding SPIE 2183, pp. 155–163, February 1994
6. Dorrity, J.L., Vachtsevanos, G.: On-line defect detection for weaving systems. In:
Proceeding IEEE Annual Technical Conference Textile, Fiber, and Film Industry, pp. 1–
6, May 1996
7. Ryan G.: Rosandich: Intelligent Visual Inspection, Chapman & Hall, London (U.K.) (1997)
8. Batchelor, B.G.: Lighting and viewing techniques. In: Batchelor, B.G., Hill, D.A., Hodgson,
D.C. (eds) Automated Visual Inspection. IFS and North Holland (1985)
9. Roberts, J.W., Rose, S.D., Jullian, G., Nicholas, L., Jenkins, P.T., Chamberlin, S.G.,
Maroscher, G., Mantha, R., Litwiller, D.J.: A PC-based real time defect imaging system for
high speed web inspection. In:Proceeding SPIE 1907, pp. 164–176 (1993)
10. Bayer, H.A.: Performance analysis of CCD-cameras for industrial inspection. In: Proceed-
ing SPIE 1989, pp. 40–49 (1993)
11. Cho, C., Chung, B., Park, M.: Development of real-time vision-based fabric inspection
system. IEEE Trans. Ind. Electron. 52(4), 1073–1079 (2005)
12. Kumar, A.: Computer-vision-based fabric defect detection: a survey. IEEE Trans. Ind.
Electron. 55(1), 348–363 (2008)
13. Ngana, H., Panga, G., Yung, N.: Automated fabric defect detection a review. Image Visi.
Comput. 29(7), 442–458 (2011)
14. Smith, B.: Making war on defects. IEEE Spectr. 30(9), 43–47 (1993)
15. Fernandez, C., Fernandez, S., Campoy P., Aracil R.: On-line texture analysis for flat
products inspection. neural nets implementation. In: Proceedings of 20th IEEE International
Conference on Industrial Electronics, Control and Instrumentation, vol. 2, pp. 867–872
(1994)
16. Ozdemir S., Ercil A.: Markov random fields and Karhunen-Loeve transforms for defect
inspection of textile products. In: IEEE Conference on Emerging Technologies and Factory
Automation, vol. 2, pp. 697–703 (1996)
 12 T. Mahmud et al.

17. Bodnarova A., Williams J. A., Bennamoun M., Kubik K. Optimal textural features for flaw
Author Proof

detection in textile materials. In: Proceedings of the IEEE TENCON 1997 Conference,
Brisbane, Australia, pp. 307–310 (1997)
18. Gong, Y.N.: Study on image analysis of fabric defects. Ph.D. dissertation, China Textile
University, Shanghai China (1999)
19. Zhang, Y.F., Bresee, R.R.: Fabric defect detection and classification using image analysis.
Text. Res. J. 65(1), 1–9 (1995)
20. Nickolay, B., Schicktanz, K., Schmalfub, H.: Automatic fabric inspection– utopia or reality.
Trans. Melliand Textilberichte 1, 33–37 (1993)
21. Habib, M.T., Rokonuzzaman, M.: A set of geometric features for neural network-based
textile defect classification, ISRN Artif. Intell. 2012, Article ID 643473, p. 16 (2012)
22. Saeidi, R.D., Latifi, M., Najar, S.S., Ghazi Saeidi, A.: Computer Vision-Aided Fabric
Inspection System For On-Circular Knitting Machine, Text. Res. J. 75(6), 492–497 (2005)
23. Islam, M.A., Akhter, S., Mursalin, T.E.: Automated textile defect recognition system using
computer vision and artificial neural networks. In: Proceedings World Academy of Science,
Engineering and Technology, vol. 13, pp. 1–7, May 2006
24. Murino, V., Bicego, M., Rossi, I.A.: Statistical classification of raw textile defects. In: 17th
International Conference on Pattern Recognition (ICPR 2004), ICPR, vol. 4, pp. 311–314
(2004)
25. Karayiannis, Y.A., Stojanovic, R., Mitropoulos, P., Koulamas, C., Stouraitis, T., Koubias,
S., Papadopoulos, G.: Defect detection and classification on web textile fabric using multi
resolution decomposition and neural networks. In: Proceedings on the 6th IEEE International
Conference on Electronics, Circuits and Systems, Pafos, Cyprus, pp. 765–768, September
1999
26. Kumar, A.: Neural network based detection of local textile defects. Pattern Recogn. 36,
1645–1659 (2003)
27. Kuo, C.F.J., Lee, C.-J.: A back-propagation neural network for recognizing fabric defects.
Text. Res. J. 73(2), 147–151 (2003)
28. Mitropoulos, P., Koulamas, C., Stojanovic, R., Koubias, S., Papadopoulos, G., Karayiannis,
G.: Real-time vision system for defect detection and neural classification of web textile
fabric. In: Proceedings SPIE, vol. 3652, San Jose, California, pp. 59–69, January 1999
29. Shady, E., Gowayed, Y., Abouiiana, M., Youssef, S., Pastore, C.: Detection and
classification of defects in knitted fabric structures. Text. Res. J. 76(4), 295–300 (2006)
30. Campbell, J.G.,. Fraley, C., Stanford, D., Murtagh, F., Raftery, A.E.: Model-based methods
for textile fault detection, Int. J. Imaging Syst. Technol. 10(4), 339–346, July 1999
31. Islam, M.A., Akhter, S., Mursalin, T.E., Amin, M.A.: A suitable neural network to detect
textile defects. Neural Inf. Process. 4233, 430–438. Springer, October 2006
32. Habib, M.T., Rokonuzzaman, M.: Distinguishing feature selection for fabric defect
classification using neural network. J. Multimedia 6 (5), 416–424, October 2011
33. TILDA Textile texture database, texture analysis working group of DFG. http://lmb.
informatik.unifreiburg.de
34. Srinivasan, K., Dastor, P. H., Radhakrishnaihan, P., Jayaraman, S.: FDAS: a knowledge-
based frame detection work for analysis of defects in woven textile structures, J. Text. Inst.
83(3), 431–447 (1992)
35. Rao Ananthavaram, R.K., Srinivasa, Rao O., Krishna P.M.H.M.: Automatic defect detection
of patterned fabric by using RB method and independent component analysis. Int.
J. Comput. Appl. 39(18), 52–56 (2012)
36. Sengottuvelan, P., Wahi, A., Shanmugam, A.: Automatic fault analysis of textile fabric using
imaging systems. Res. J. Appl. Sci. 3(1), 26–31 (2008)
 Fabric Defect Detection System 13

37. Abdi. H., Williams, L.J.: Principal component analysis. Wiley Interdisciplinary Rev.
Author Proof

Comput. Stat. 2 (4), 433–459 (2010). https://doi.org/10.1002/wics.101

38. Kumar, T., Sahoo, G.: Novel method of edge detection using cellular automata. Int.
J. Comput. Appl. 9(4), 38–44 (2010)
39. Zhu, Q.: Efficient evaluations of edge connectivity and width uniformity. Image Vis.
Comput. 14, 21–34 (1996)
40. Senthilkumaran. N., Rajesh, R.: Edge detection techniques for image segmentation – a
survey of soft computing approaches. Int. J. Recent Trends Eng. 1(2), 250–254 (2009)
41. Rizon, M., Hashim, M.F., Saad, P., Yaacob, S.: Face recognition using eigen faces and
neural networks. Am. J. Appl. Sci. 2(6), 1872–1875 (2006)
42. Rosten, E., Porter, R., Drummond,T.: FASTER and better: a machine learning approach to
corner detection, IEEE Trans Pattern Anal Mach Intell. 32, 105–119 (2010)
43. Wikipedia, Corner Detection. http://en.wikipedia.org/wiki/Corner_detection. Accessed 16
March 2011
44. Chang, J.Y., Chen, J.L.: Automated facial expression recognition system using neural
networks. J. Chin. Inst. Eng. 24(3), 345–356 (2001)
45. Jianli, L., Baoqi, Z.: Identification of fabric defects based on discrete wavelet transform and
back-propagation neural network. J. Text. Inst. 98(4), 355–362 (2007)
46. Tamnun, M.E., Fajrana, Z.E., Ahmed, R.I.: Fabric defect inspection system using neural
network and microcontroller. J. Theor. Appl. Inf. Technol. 4(7) (2008) AQ3
47. Bhanumati, P., Nasira, G.M.: Fabric inspection system using artificial neural network. Int.
J. Comput. Eng. 2(5), 20–27 May 2012

View publication stats