Texture is everywhere in our everyday life. When going to a clothing store, we can differentiate two white shirts, which one is made of silk or linen without having to touch them. The ripeness of a fruit can also be determined by its texture. Beyond these daily tasks, texture is also used as cues for quality control in manufacturing industries as well as in the medical field. In the wood processing industry, texture information has been used to evaluate wood surface quality, to further detect defects. It has also been used to aid medical doctors in providing cancer diagnosis.
Our human visual system is limited to the visible range of the electromagnetic spectrum; we can only see the colors of an object or surface. Just by looking, we cannot tell whether a pot is hot. We also do not have the capability to judge which log of wood has more water content. This is because the information about heat and water content lies in the infrared range of the electromagnetic spectrum, thus invisible to us. Hyperspectral imaging (HSI) is an imaging technology that can capture such information and at a much more detailed resolution.
Despite the cost and complexity of a hyperspectral image acquisition, the use of HSI thrives in many fields due to its potential in providing highly accurate measurement and analysis results. However, the current ways of exploiting hyperspectral images, including their texture analysis, do not live up to the potential they offer since they have not been correctly treated as measurement data. In this project, we aim to enforce metrology principles into the texture analysis of hyperspectral images, in which a hyperspectral image will thus be treated as a measurement through its entire chain of processing. Concretely, the project goal is to produce not only metrological image processing tools for hyperspectral images, but also a set of quality assessment protocols enabling to quantify bias, uncertainty, and other metrological units of the processed data.