Machine Learning for Volumetric Data Analysis of Bread Dough: Meeting the Synchrotron
Challenge
A thesis submitted in fulfilment of the requirements for the degree of Master of Engineering
Salah Mohammed Abdulrahman Ali
Bachelor of Engineering (Honours), IIUM
School of Engineering
College of Science, Technology, Engineering and Maths
RMIT University
October 2020
i
Declaration
I certify that except where due acknowledgement has been made, the work is that of the
author alone; the work has not been submitted previously, in whole or in part, to qualify
for any other academic award; the content of the thesis is the result of work which has
been carried out since the official commencement date of the approved research pro-
gram; any editorial work, paid or unpaid, carried out by a third party is acknowledged;
and, ethics procedures and guidelines have been followed. I acknowledge the support I
have received for my research through the provision of an Australian Government Re-
search Training Program Scholarship.
Salah Mohammed Abdulrahman Ali
01 October 2020
ii
Acknowledgement
I would like to express my deepest gratitude to my supervisors Prof. Alireza Bab-
Hadiashar, Amirali Khodadadian Gostar, Dr. Sherry Mayo, and Dr. Ruwan Tennakoon
for their continuous encouragement, supervision, support, and more importantly, for
providing an amazing environment to achieve this milestone in my life. You were such
a great mentors for me, and I learnt a lot from you.
A special thank to CSIRO Manufacturing team, Dr Sherry Mayo, Dr Aaron Seeber,
Darren Thompson, and the rest of the team for helping me during my candidature. You
always offer me with all the resources, time and advice which helped in the development
of this work.
My sincere thanks go to my colleagues in Intelligent Automation Research Group at
RMIT, for their value feedback in the daily discussions.
My special appreciation and gratitude go to my parents, lovely wife, Heyam, my little
one Yamin, and my brother Ashraf. You were there for me whenever I needed you, and
you always provided me with the encouragement to continue during all my study.
Finally, I acknowledge the support I have received for my research through the provision
of an Australian Government Research Training Program Scholarship. This scholarship
makes achieve this project possible.
Contents
Declaration i
Acknowledgement ii
Contents iii
List of Figures v
List of Tables viii
Symbols ix
Abstract x
1 Introduction 1
1.1 Motivation ................................. 1
1.2 Problem Statement ............................ 3
1.3 Scope ................................... 3
1.4 Research Objectives ............................ 3
1.5 Research Questions ............................ 4
1.5.1 Research Question 1 . . . . . . . . . . . . . . . . . . . . . . . 4
1.5.2 Research Question 2 . . . . . . . . . . . . . . . . . . . . . . . 4
1.5.3 Research Question 3 . . . . . . . . . . . . . . . . . . . . . . . 4
1.6 Research Methodology .......................... 5
1.7 Thesis Outline ............................... 6
2 Literature Review 7
2.1 Introduction ................................ 7
2.2 Synchrotron Based Micro CT System .................. 7
2.2.1 Components of Micro CT System ................ 7
2.2.2 Synchrotron Based Micro CT System .............. 8
2.2.3 Imaging and Medical BeamLine (IMBL) ............ 10
2.3 Bread Studies ............................... 11
2.3.1 Image Processing Techniques for Bread Studies ......... 11
2.4 Machine Learning ............................. 13
2.4.1 Machine Learning Overview . . . . . . . . . . . . . . . . . . . 13
iii
Contents iv
2.4.2 Supervised Learning . . . . . . . . . . . . . . . . . . . . . . . 14
2.4.3 Unsupervised Learning . . . . . . . . . . . . . . . . . . . . . . 18
2.4.4 Semi-supervised Learning . . . . . . . . . . . . . . . . . . . . 19
2.5 Deep Learning .............................. 21
2.5.1 Deep Learning Overview . . . . . . . . . . . . . . . . . . . . . 21
2.5.2 Types of Neural Networks . . . . . . . . . . . . . . . . . . . . 21
2.5.3 Application of deep learning on Computer vision ........ 25
2.5.3.1 Classifications . . . . . . . . . . . . . . . . . . . . . 25
2.5.3.2 Object Detection . . . . . . . . . . . . . . . . . . . . 26
2.5.3.3 Segmentation . . . . . . . . . . . . . . . . . . . . . 27
2.5.4 Application of Deep Learning to analyse porous materials . . . 29
3 Data Acquisition 30
3.1 Introduction ................................ 30
3.2 Data Collection .............................. 30
3.3 Bread dough baking process . . . . . . . . . . . . . . . . . . . . . . . 31
3.4 Reconstruction Micro-CT images . . . . . . . . . . . . . . . . . . . . . 32
3.5 Representative elementary volume (REV) analysis ........... 36
3.6 Preparation of testing datasets . . . . . . . . . . . . . . . . . . . . . . 37
3.6.1 Analysis of testing datasets . . . . . . . . . . . . . . . . . . . . 40
3.7 Generating synthetic training data . . . . . . . . . . . . . . . . . . . . 40
4 Segmentation of Porous Bread dough 45
4.1 Introduction ................................ 45
4.2 Segmentation based on Threshold . . . . . . . . . . . . . . . . . . . . 45
4.3 Segmentation based on U Net . . . . . . . . . . . . . . . . . . . . . . . 47
4.3.1 Model Architecture . . . . . . . . . . . . . . . . . . . . . . . . 47
4.3.2 Model Training .......................... 47
4.4 Results and Discussion .......................... 48
4.4.1 Comparison of U-net and Otsu segmentation .......... 48
5 Extracting Structural Information 51
5.1 Introduction ................................ 51
5.2 Qualitative Metrics ............................ 51
5.3 Snow Algorithm .............................. 52
5.4 Results ................................... 52
6 Conclusion and future work 55
6.1 Conclusion ................................ 55
6.2 Future work ................................ 57
Bibliography 58
Appendix 64