Rising concerns over the supply of oil and gas and the impacts of climate change are increasing demand for renewable energy sources to relieve the current dependence on fossil fuels. Hydrogen is among the most promising alternative sources of energy and as a result, the research and development of hydrogen fuel cells is growing rapidly worldwide. Researchers at the University of Queensland’s Centre for Microscopy and Microanalysis (CMM) are involved in a collaboration working on various aspects of fuel cell technology, including electrode characteristics, electrolyte design and information management systems to support fuel cell research and development.
At each stage of such research, whether it involves material design, performance testing or manufacturing, extensive data collection and analysis must be conducted. In particular, reliable and detailed information is needed regarding the microstructure of the fuel cell materials so that knowledge of composition can be correlated with the measurable macroscopic fuel cell properties. The FUSION (Fuel cell Understanding through Semantic Inferencing, Ontologies and Nanotechnology) project, run by the Distributed Systems Technology Centre at UQ, aims to apply semantic web technology to the cross-correlation and analysis of the large data sets that are obtained during the fuel cell workflow.
One of the components of this system, then, is the extraction and analysis of microstructural information from material sample imaging so that it can be indexed and referenced by the FUSION metadata set. The different materials produce characteristic images and a customised, automated process of image analysis is the most efficient way to access the constituent data. This project discussed here is an image analysis study of a set of electrolyte images with the purpose of developing an appropriate method of analysis using the Matlab software package.
The basics of image analysis theory and a range of processing methods have been investigated and applied to a set of test images. A summary of a part of this theory is included in this report as background to the processing methods that are presented. Progress has been made toward developing a method that is capable of automatically extracting and measuring the features in the microscopic images. The specific requirements of the images have been assessed and techniques of image enhancement, feature detection, binary processing and image measurement successfully implemented. Unfortunately, the analysis process that is the outcome of the study is not implemented with full automation and does not achieve results of sufficient accuracy for useful quantitative measurement. Further work to improve or redevelop the methods that are applied is required to accomplish these goals.
The image analysis investigations and method are described in some detail in this document and the set of results that are obtained are presented. It is hoped that this preliminary study will be of use in developing more robust and comprehensive methods of microstructural image analysis to aid the development of the fuel cell systems.