£7 

^flBRARV 
" U B I V E R S I T Y OF M O R A T U W A . SR I I A N « 

— — M O R A T U W A 

A Neural Network Approach to Classification of Textile 

Defects 

J.T.S. Priyadarshani 

08/10036 

University of Moratuwa 

O 0 4 f 11 

102481 

Dissertation submitted to the Faculty of Information Technology, University of 

Moratuwa for the partial fulfillment of the requirements of the of MSc. (PG Dip) in 

Information Technology /O^tyS f 

February 201 

102481 

Lb/ DO*/ /9/20/X 



Declaration 
I declare this thesis is my own work and has not been submitted in any form for 

another degree or diploma at any other university or other institution of tertiary 

education. Information derived from the published or unpublished work of others has 

been acknowledged in the text and a list of references is given. 

JTS Priyadarshani 

Name of Student Signature of Student 

Date: |3|c7f>Uo// 1 

Supervised by 

Prof. AS Karunananda 

Name of Supervisor(s) Signature of Supervisor(s) 

Date: W/C,*/**' 

ii 



Dedication 

This Outcome is lovingly dedicated to Thaththa, Amma and Upul. 

iii 



Acknowledgments 
While I am preparing this report, I turned back to convey my gratitude to many people 

for their valuable time spend towards me. This research will be benefited in many 

ways through their generosity and opinions which are expressed are not to be 

interpreted as held by those whose courtesy and helpfulness are here acknowledged. 

First, I wish to sincerely thank the research supervisor, Prof. A S Karunananda, Dean, 

Faculty of Information Technology, for guiding me to complete this valuable research 

project. 

Secondly, I wish to express my sincere gratitude to the senior lecturers, Mr. Saminda 

Premaratne in Department of Information Technology and Dr Ranga Rodrigo 

Department of Electronic and Telecommunication Engineering for their valuable 

advices on the technical matters. Further I would like to express my heartfelt 

appreciation to Mr. Sahan Ranasinghe, Department of Electrical and Information 

Engineering in the University of Ruhuna and Rasika Jayewardena, Instructor in the 

Department of Electrical Engineering for spending their valuable time in sharing their 

knowledge with me. 

In fact, I would like to thank, Mr. Upul Dharmadasa and Ms. Nilupa Liyanage, Indika 

Samarasinghe and Dilhan Thilakarathne for assisting me in different ways in making 

this project a success in the initial stage. 

In addition, I wish to convey my heartiest gratitude to Mourice Jochim, Product 

Dvelopment manager of Penguine Sportswear, Mr. Chandika Dharmadasa, Quality 

Executive, Brandix Casualwear, Mr. Mr. Sisira Bandaranayake, Group Sample Room 

Manager, Emjeyi Apparel Ltd, Mr. Kapila Jayesundara, Quality manager in MAS 

Casual Line, Mr. Lahiru Ranasinghe, Metropolitan Ltd by providing me in requiring 

hardware and software. 

Last but not least I convey my heartiest gratitude to my dearest parents and all the 

others who guided me even in a word to make this outcome a perfectly one. 

Finally I would like to thank Ms. Judith Labrooy for proof reading this report within 

very short period of time. 

Thank you all! 

JTS Priyadarshani 

Faculty of IT 

iv 



Abstract 
All textile industries aim to produce competitive textiles. The competition 

enhancement depends mainly on productivity and quality of the textiles produced by 

each industry. Especially in the least developed countries like Sri Lanka, India, 

Bangladesh where textile is one of the main incomes of economy, still using manual 

quality control techniques to identify defects in their textile products. Therefore this 

study carries out to classify defects in textiles with minimum intervention of human 

being using artificial neural network technology and some principles of image 

processing techniques. 

My approach is to identify defects in textile comprises three steps, namely, image 

processing, image classification and producing the output as a classification chart. I 

have used the standard techniques of image processing; while the image classification 

is handled by an ANN trained in the supervised mode with the aid of back 

propagation training algorithm. I have selected three major defect types of defects 

from textile quality process. They are cut hole damages, open seam of the garment 

and pen marks. Input data is a 185220x30 matrix and it is representing 30 samples of 

185220 elements. The output of the system will be an image file with a defect 

classification chart. In fact, basically this system is able to classify defects in to three 

categories. 

This system mainly is divided in to three sections namely; image processing module, 

neural network module and output generator module. But the system design is 

separated in to two parts. The first part of the research processes the images to fit for 

the input layer of the neural network. The second part uses the input set to classify the 

defects and adapts the neural net. 

The prototype develop in this project have been trained to identify cut hole damages, 

open seam and pen mark of the garments by considering 300 images. A concatenate 

image, which contained 300 samples of each defect type, was trained and tested. The 

accuracy rate of recognizing the numbers and symbols were 77%. However this 

accuracy rate is very high. 

Using the neural network technology and back propagation learning algorithm, to 

classify textile defects were successful even there was a high probability to 

misclassify different types of defect images with different range of severity rates. 

Keywords: Neural network, Back propagation learning algorithm 



Contents 

Chapter 1 Introduction 1 

1.1 Introduction 1 

1.2 Textile Quality Control System 1 

1.3 Background and Motivation 3 

1.4 Aim and Objectives 3 

1.5 Towards the Technology and Solution 4 

1.6 Resource Requirement 4 

1.7 Structure of the Dissertation 4 

1.8 Summary 5 

Chapter 2 Review of Textile Defects Classification Systems 6 

2.1 Introduction 6 

2.2 Current Approaches for Textile Defects Recognition and classification Systems 6 

2.3 Discussion on Others' Approaches 8 

2.4 Summary 8 

Chapter 3 Principles of Image Processing and Artificial Neural Networks 10 

3.1 Introduction 10 

3.2 Review of Available Technologies 10 

3.2.1 Principle Component Analysis (PCA) 10 

3.2.1.1 Objectives of principal component analysis 10 

3.2.1.2 Limitations of PCA 11 

3.2.2 Statistical Models 11 

3.2.3 Artificial Neural Network (ANN) 11 

3.2.3.1 Why use neural networks? 11 

3.3 Image Processing 12 

3.4 Artificial Neural Network 13 

3.4.1 Network layers 15 

vi 

Page 



3.4.2 Types of Neural Networks 15 

vn 

3.4.2.1 Multi-Layer Feed -forward Neural Networks 16 

3.4.2.2. Feedback networks 17 

3.4.2.3 Back Propagation Algorithm 17 

3.5 Summary 19 

Chapter 4 Approach 20 

4.1 Introduction 20 

4.2 Textile Defects classification Using Neural Network 20 

4.2.1 Input 20 

4.2.2 Output 21 

4.2.3 User 21 

4.2.4 Processes 22 

4.2.5 Features 22 

4.3 Summary 22 

Chapter 5 Analysis and Design 23 

5.1 Introduction 23 

5.2 Top Level Architecture 23 

5.3 Image Processing Module 24 

5.3.1 Normalization 24 

5.3.2 Thresholding 24 

5.3.3 Segmentation 25 

5.3.4 Feature Extraction 25 

5.4 Artificial Neural Network Module 25 

5.5 Output Generator Module 26 

5.6 Summary 26 

Chapter 6 Implementation of the Textile Defects Classification System 28 

6.1 Introduction 28 



6.2 Creating the Textile Defects Classification System 

viii 

6.3 Preparing Training and Testing Data Set 

6.3.1 Training 

6.3.2 Validation 

6.3.3 Testing 

6.3.4 Concatenate images; 

6.4 Normalization and Reading Pixel Value of the Image 

6.5 Neural Network Architecture 

6.6 Training and Learning Mechanism 

6.6.1 Setting the Weights 

6.6.2 Backpropagation Algorithm 

6.6.3 Training of Network 

6.6.4 Training performance 

6.7 Simulation Results 

6.8 Error Rate Estimation 

6.9 Output Generator Module 

6.10 Summary 

Chapter 7 Evaluation of the Textiles Defects Classification Syst< 

7.1 Introduction 

7.2 Evaluation of Textile Defects Classification System 

7.3 Summary 

Chapter 8 Conclusion and Further Works 

8.1 Introduction 

8.2 Conclusion 

8.3 Limitations 

8.4 Further Improvements 

8.5 Summary 



References 

ix 

Appendix A - Use Case Diagram for Recognition of Textile Defects 

Appendix B - Source Code of Create a Neural Network 

Appendix C - Source Code of GUI Design 



List of Figures 
Page No. 

Figure 3.1: Multi Layer Artificial Neural Networks 15 

Figure 3.2: Three layers feed - forward neural net 16 

Figure 3.3: Processing unit element 17 

Figure 3.4: Multilayer feedback ANN 17 

Figure 3.5: Forward and Backward propagation 18 

Figure 5.1: Top Level Architecture of the System 23 

Figure 6.1: System Design of Textile Defects Detector 29 

Figure 6.2: Normalization (55 X 55 Open seam) 32 

Figure 6.3: Neural Network Architecture 33 

Figure 6.4: Flow Chart of Training the Network 36 

Figure 6.5: Training the Network with nftool in Matlab 37 

Figure 6.6: Training Performance 38 

Figure A. l : Use case Diagram for Defect Recognition 50 

x 



List of Tables 
Table 1.1: Discussions on Others' Approaches 

Table 6.1: Results of Classification 

Table 6.2: Output Layer Target Neuron Values 

xi