Virtual leaves: from data to surfaces and the steps in-between

Overview

Like all industries, agriculture is benefiting from the data and computing revolution. Using hand-held scanners, CT scanners, or other technologies, we can acquire data sets that represent real leaves of agricultural crops, e.g. wheat. Using this data, and performing many intermediate steps, we can build virtual leaf surfaces that can be used in computer models to perform simulations of droplet impactions, spreading, evaporation, and other phenomena of interest to the industry.

This project concerns the 'many intermediate steps', for which there are countless interesting and yet-unexplored pathways that could utilise and combine techniques from applied mathematics, optimisation, data science, and computer science.

Research activities

Depending on student experience and interests, research activities include:

• linear algebra problems of surface fitting to noisy data
• optimisation of parameters in surface fitting
• dimensionality reduction techniques and nonlinear projections
• surface triangulation from implicit volumetric data
• image quilting or related dynamic programming technologies to seamlessly extend microstructure features on surfaces
• machine learning techniques to learn microstructure surface features from CT data
• computer graphics techniques for synthetic microstructure surface generation.

This project forms part of a larger ARC Linkage Project. As such, you will have the opportunity if you wish to present your work as part of the research group, gaining experience with working as a member of a much larger project.

Outcomes

Improvements in existing pipelines going from data to surfaces. These improvements could ultimately find their way into the three-dimensional simulation software being developed by the larger research group for the agricultural industry.

Depending on the novelty and extent of the project outcomes, a scientific publication could arise from this work.

Skills and experience

This project is suitable if you have a background in mathematics, data science, and computer science backgrounds. Depending on interest and background, there are many novel aspects of this project that can be explored. You're encouraged to apply if you have a background in:

• linear algebra
• optimisation
• dynamic programming
• machine learning
• graph theory
• computer graphics.

Contact

Contact the supervisor for more information.