Student Projects

Below is a selection of example student projects which may be undertaken as vacation projects, honours projects, or extended to a Masters or PhD level project. A vacation project at the PBRC is a great way to get research experience and learn valuable new skills. Different projects are available, and looking at our research page and individual staff home pages will give an idea on the breadth of topics available. For more information please contact Professor Stan Miklavcic or the PBRC staff member that you are interested in undertaking a project with.

Automatic Segmentation of 3D shapes

The purpose of this project is to implement and test a new method for the segmentation of 3D shapes that have tubular structure. Examples of such shapes are roots and shoots of plants. 3D segmentation is an important step towards the analysis of the plant structure. Several 3D segmentation methods have been already developed by the hosting research group but intended for various types of 3D shapes. The role of the student is to extend these methods to 3D plants (root and shoot) and evaluate their performance.

The student will have the unique opportunity to work within a multi-disciplinary research centre composed of computer scientists (working on image processing and computer vision), mathematicians and plant biologists.

 Requirements

  • •          Ability to program in Matlab
  • •          Familiarity with 3D geometry
  • •          Motivation to learn and explore new concepts

Supervisors

Dr Jinhai Cai

References

1. Hamid Laga, Michela Mortara and Michela Spagnuolo. “Geometry and Context for Semantic Correspondences and Functionality Recognition in Manmade 3D Shapes. ACM Transactions on Graphics, 32(5), 2013.

2. Pankaj Kumar, Jinhai Cai, and Stanley J. Miklavcic. Improved Ellipse Fitting by Considering the Eccentricity of Data Point Sets. In IEEE International Conference on Image Processing (ICIP) 2013.

Automatic Extraction of the Vein Structure of Plant Leaves from Images

The 2D shape of a plant leaf, its margin and its vein structure are of great importance to plant scientists as it can help in distinguishing species, measuring plant health, analysing growth patterns and understanding relations between various species. While shape and margin have been extensively studied and used in classification tasks, the vein structure has seldom been explored due to the difficulty in segmenting it from images. The goal of this project is to investigate and develop techniques for automatic extraction of the leaf vein structure from 2D images and to characterize the extracted structure with descriptors for classification tasks. The output of this project will provide computational tools to plant scientists for studying evolution and evolutionary relationships among species and for modelling their continuous variability from the shape perspective.

The student will have the unique opportunity to work within a multi-disciplinary research centre composed of computer scientists (working on image processing and computer vision), mathematicians and plant biologists.

Requirements

  • Ability to program with Matlab
  • Motivation to learn and explore new concepts.

Supervisors

Dr Josh Chopin, Professor Stan Miklavcic, Dr Hamid Laga

References

1. Hamid Laga, Sebastian Kurtek, Anuj Srivastava, Stanley J. Miklavcic. Statistical Shape Models of Plant Leaves. International Conference on Image Processing (ICIP) 2013.

2. Hamid Laga, Sebastian Kurtek, Anuj Srivastava, Mahmood Golzarian, and Stanley J. Miklavcic. A Riemannian Elastic Metric for Shape-based Plant Leaf Classification. IEEE International Conference on Digital Image Computing (DICTA), pp. 1-7, 2012.

Controlled drug delivery with multi-layered tablets

 In order to control the effects of many illnesses, the steady release of a drug into the body is necessary. However, a drug (in tablet form) is usually only administered a few times per day, with each dosage leading to a dramatic increase in the drug concentration within the body, followed by a slow decrease as the drug is metabolised. Fortunately, it is now possible to manufacture layered tablets, with the drug concentration and release rate varying between the layers. This results in a more steady release of the drug into the system, and maintains it at a more constant concentration within the body. In this project, we will tackle this problem from two directions. First, we will calculate the rate of drug delivery assuming that we know the concentration of the drug within each layer and also the rate at which each layer dissolves. Second, we tackle the more difficult inverse problem: if the required drug delivery rate is known, how could you construct a tablet with a finite number of layers that would closely replicate the necessary delivery rate. (Suitable as an honours project)

Key Words: Partial differential equations

Contact person and details: Dr Bronwyn Hajek

E Bronwyn.Hajek@unisa.edu.au; T 8302 3084; URL http://people.unisa.edu.au/Bronwyn.Hajek

Creating nanopatterns by dewetting polymer brushes

Polymer brushes are polymer chains that have been grafted by one end on to a solid substrate. In the presence of a solvent, the polymer chains are stretched away from the substrate, however, if the solvent surrounding polymer brushes dries out, the polymer brushes collapse onto the substrate in a compact layer. Molecular dynamics simulations have shown that as the brushes collapse, they can form nanopatterns on the substrate, with the type of pattern depending on the grafting density and the amount of solvent. In this project, we will test the robustness of these conclusions using partial differential equations, in much the same way as Murray describes the patterning on mammalian coats (eg leopards and zebras). (Suitable as an honours project)

Key Words: Partial differential equations

References:

  1. T Lee, SC Hendy, C Neto, Tunable nanopatterns via the constrained dewetting of polymer brushes, Macromolecules, 2013, 46(15):6326-6335
  2. JD Murray, Mathematical Biology II: Spatial Models and Biomedical Applications, Springer-Verlag, Berlin, 2003

Contact person and details: Dr Bronwyn Hajek, Dr Marta Krasowska

E Bronwyn.Hajek@unisa.edu.au; T 8302 3084; URL http://people.unisa.edu.au/Bronwyn.Hajek

Deep learning on plant image analysis

Deep learning methods have experienced an immense growth in interest from the machine learning and computer vision community in recent decade. This is because it has been successfully applied on a large amount of challenge tasks, such as face recognition, medical image analysis and automatic language translation. However, there is little research on plant image analysis using the deep learning methods. This project is dedicated to developing an automatic deep learning-based algorithm to have the biological traits (such as leaf coverage, spikes detection and tiller numbers) from the plant images. (suitable for honours or MSc student)

Key Words: Deep learning, machine learning, image processing, image-based plant phenotyping

Contact person and details: Dr Zohaib Khan, Dr Jinhai Cai

E Zohaib.Khan@unisa.edu.au; URL http://people.unisa.edu.au/Zohaib.Khan

Mathematical models for microelectromechanical machines

Microelectromechanical machines are increasingly being used as sensors and actuators. At present, their performance is limited due to issues with contamination and friction. In this project, we will develop a mathematical model to investigate the mechanisms which govern the interactions in these devices. In particular, we will model the surface forces within these devices and investigate the use of liquid lubricants, combined with specially designed coatings. (Suitable as PhD project)

Key Words: Applied math modelling, Physical Chemistry

Contact person and details: Dr Bronwyn Hajek, Professor Jim Hill, Dr Marta Krasowska, Associate Professor David Beattie

E Bronwyn.Hajek@unisa.edu.au; T 8302 3084; URL http://people.unisa.edu.au/Bronwyn.Hajek

Modelling of fluid and solute transport in non-uniform, periodic capillaries

The flow of fluids and transport of suspending particles in capillaries has attracted a lot of experimental and theoretical interest in recent years. The interest is partly inspired by the potential for commercial exploitation in the area of microfluidics and nanofluidics applications in chemical and pharmaceutical industries. However, inspiration also comes from a desire to understand a range of natural phenomena, such as arise in plants. We have an interest in extending our recent efforts to model fluid flow and particle transport in periodic tubes to more general tube conditions, on the one hand, and considering more detailed (perturbation or asymptotic) analyses in simpler cases, on the other. (Suitable as PhD project)

Key Words: Applied mathematics/mathematical modelling

References:

  1. N. Islam, B. Bradshaw-Hajek, S.J. Miklavcic, L.R. White (2015) “The onset of recirculation flow in periodic capillaries: geometric effects”, European Journal of Mechanics - B/Fluids, 53, pp119-128.
  2. N. Islam, S.J. Miklavcic, B. Bradshaw-Hajek, L.R. White (2017) “Convective and diffusive effects on particle transport in asymmetric periodic capillaries”, PLoS One 12(8): e0183127.

Contact person and details: Professor Stanley Miklavcic, Dr Bronwyn Hajek

E Stan.Miklavcic@unisa.edu.au; T 8302 3788; URL http://people.unisa.edu.au/Stan.Miklavcic

Modelling of salt and water transport in plants

Abiotic stresses such as high salt levels in soils can severely affect cereal crop health, development and grain yield. Currently, high salinity affects two-thirds of Australian cereal crops. To increase plant salinity tolerance it is necessary to manipulate the transport of ions (e.g., sodium and chloride) through a plant. However, this requires knowledge about how ion transport through a plant occurs. In particular, it is necessary to identify the key points in this transport pathway to target in order to generate a salt-tolerant cereal variety. For example, is targeting the initial influx of ions from the soil the best method for increasing plant salinity tolerance, or should more effort be directed towards increasing the compartmentalization of ions in the shoot? None of the existing models of water and solute transport in plants are currently suitable for analysing the transport of ions. This project aims to develop detailed mathematical models of water and solute transport through plant organs and tissues, which will be compared with physiological measurements of fluxes and accumulated ion concentrations. The overall aim is to aid understanding of the biophysical mechanisms and processes responsible for increasing plant salinity tolerance. It is envisioned that the results will help guide plant geneticists and plant breeders in their search for specific genetic traits that enhance a plant's ability to tolerate salinity. (Suitable as PhD project)

Key Words: Applied mathematics/mathematical modelling

References:

  1. K. Foster and S.J. Miklavcic. “Mathematical modelling of the uptake and transport of salt in plant roots”, J. Theoretical Biology, 336, pp132-143.
  2. K. Foster and S.J. Miklavcic. “On the competitive uptake and transport of ions through differentiated root tissues”, J. Theoretical Biology, 340, pp1-10.
  3. K. Foster and S.J. Miklavcic. “Toward a biophysical understanding of the salt stress response of individual plant cells”, J. Theoretical Biology, 385, pp130-142.

Contact person and details: Professor Stanley Miklavcic

E Stan.Miklavcic@unisa.edu.au; T 8302 3788; URL http://people.unisa.edu.au/Stan.Miklavcic

Modelling of surface forces in ionic liquids

Ionic liquids or molten salts are very highly concentrated salts in a fluid state. Such systems feature prominently in many chemical industry processes. However, their behavior has not been completely nor adequately quantified. In particular, how ionic liquids influence the interaction between macroscopic surfaces is not known, with conflicting experimental studies confusing the picture. This is a theoretical project aimed at developing a mathematical model to describe the forces between macroscopic surfaces in the presence of an intervening ionic liquid. The project involves the application of  advanced statistical mechanical models to help understand how ionic liquids influence the forces. We shall compare the results with published data as well as new in-house surface force measurements. (Suitable as PhD project)

Key Words: Applied mathematics/mathematical modelling

Contact person and details: Professor Stanley Miklavcic, Dr Jason Connor

E Stan.Miklavcic@unisa.edu.au; T 8302 3788; URL http://people.unisa.edu.au/Stan.Miklavcic

Plant Segmentation Using Colour Pixel Classification: Analysis and Comparison

This project will involve labelling of plant image data into foreground and background.  A study will then be conducted into three important aspects of the colour pixel classification approach to plant segmentation:  colour representation, colour quantization, and classification algorithms.  The colour spaces to be studied will be RGB, HSV, YCbCr, CIE-lab, and Chromacity and intensity spaces.  The behaviour of segmentation with respect to the colour spaces and the separation of chromacity and intensity spaces will be analysed in different segmentation algorithms.  The effect of different levels of quantization in different colour spaces is also to be studied.  The segmentation algorithms to be studied are Bayesian classification with Gaussian Mixture modelling, Support vector machines, Neural networks, thresholding, and some commonly used heuristics.  This study will lead to the study of some research ideas on phototyping study of plant senescence by imaging of plants.

Supervisor

Dr Jinhai Cai, Dr Josh Chopin, Professor Stan Miklavcic

Root Phenotyping

 Abstract:  Phenotyping of plant root growth by image processing and analysis of plant root images.  In this research automated algorithms for localization and identification of root features in image sequences of different cereal plants will be developed. The plant roots are grown in transparent gellan gum medium and imaged daily at different rotation angles. Feature extraction and feature selection techniques are employed to extract phenotyping features in root images. Features are matched and tracked across spatially separated images to extract 3D information of the phenotyping feature. Time series of 3D features are obtained by temporally tracking the features.  The project will involve development of

  • Feature extraction algorithms for plant root images
  • Feature selection and classification algorithm
  • 3D thinning algorithms

Supervisor

Dr Jinhai Cai, Dr Josh Chopin, Professor Stan Miklavcic

Sequential data analysis by integrating hidden Markov modelling with domain knowledge

Hidden Markov Models (HMMs) are statistical models of sequential data that have been used successfully in many applications in artificial intelligence, pattern recognition and modelling of gene sequences. This project aims at developing new statistics modelling approach to integrate conventional HMMs with experts’ prior knowledge (domain knowledge) to improve the capacity and the accuracy of the HMMs.

Previous works on HMMs focus on how to capture the statistic information from the sequential data and the relationships between events in time sequences. In this approach, we will develop new structure for HMMs, likely the multilayered and coupled structure, to represent domain knowledge, structure information as well statistic information into individual models. The developed novel HMMs will be applied to biology and health science. (Suitable as PhD project)

Key Words: Image processing, computer vision and machine learning

References:

  1. J. Cai and Z.Q. Liu, “Integration of structural and statistical information for unconstrained handwritten numeral recognition,” Pattern Analysis and Machine Intelligence, IEEE Transactions on 21 (3), 263-270.
  2. J. Cai and Z.Q. Liu, “Pattern recognition using Markov random field models”, Pattern Recognition 35 (3), 725-733, 2002.
  3. J. Cai, D. Ee, R. Smith, “Image Retrieval Using Circular Hidden Markov Models with a Garbage State”, IVCNZ 2007, 115-120.
  4. J. Cai, “Enhanced HMM for the Recognition of Sigma70 Promoters in Escherichia coli”, Digital Image Computing: Techniques and Applications (DICTA), 2008, 46-51.

Contact person and details: Dr Jinhai Cai, Professor Stan Miklavcic, Dr Hamid Laga (Murdoch University)

E Jinhai.Cai@unisa.edu.au; T 8302 5533; URL http://people.unisa.edu.au/Jinhai.Cai

Symmetry methods for nonlinear partial differential equations

 In this project we will apply a number of symmetry methods to determine solutions of some well-known nonlinear ordinary and partial differential equations, such as those arising in general relativity, acoustics, finance, environmental and biological situations, and industrial processes. In this project we will exploit the use of Lie point (classical) symmetry analysis and other modern approaches to solving PDEs. Lie point symmetry analysis provides a powerful method for finding groups of transformations which enables one to transform the ODE or PDE to an equivalent equation of simpler form. In this way, exact analytical and numerical solutions may be found.

(Suitable as PhD project)

Key Words: Applied Mathematics, partial differential equations

Contact person and details: Dr Bronwyn Hajek, Professor Jim Hill

E Bronwyn.Hajek@unisa.edu.au; T 8302 3084; URL  http://people.unisa.edu.au/Bronwyn.Hajek

Areas of study and research

+ Click to minimise