Hyperspectral imaging technology for material classification

Hyperspectral imaging (HSI) can be used as a fast and non-contact tool for quality inspection or condition assessement of material/product surfaces. The application domain of HSI is very broad. Examples of possible use-cases are the autonomous detection of corrosion and determining the condition of coatings.

The hyperspectral imaging technology is based on the fact that reflectance from a material in the electromagnetic radiation is a function of wavelength and chemical-mineralogical properties of that material. Hyperspectral imaging has applications in many different fields, from mineralogy, marine studies and forestry to agriculture, food industry and medicine. Corrosion and coating problems can also be investigated using HSI technology.

The University of Antwerp in collaboration with Engie Laborelec performed a feasibility study on some test samples, including a part of a gas turbine combustion chamber. The piece was coated with two different coatings, an MCrAIY coating and a TBC (thermal barrier coating) layer, but delamination was observed and the base metal was exposed. As both the TBC coating and the base metal have a brown colour, they are difficult to be distinguished visually.

The piece was scanned with two different cameras: a visible near-infrared photon-focus camera (NIR range: 665 nm to 975 nm) and a hyperspectral camera in short-wave infrared range (SWIR range: 900 nm to 1.700 nm).  Although the spatial resolution of the photon-focus NIR camera is higher, the spectral behaviour of materials in the visible range are very similar and hence the materials can not be distinguished from each other. The SWIR camera has much higher spectral resolution (224 continuous bands) and the spectral behaviour of each material becomes more distinctive, which enables more accurate material characterization.

 Delaminated sample provided by Engie Laborelec, scanned and analysed by University of Antwerp: spectral behaviour of 3 classes of material looks similar in visible range, while the 3 spectra in the SWIR range have distinctive behaviour, resulting in a better classification map.

Interested to learn more about innovative camera techniques for efficient assessment of coatings, corrosion and material surfaces? On 8 September you can follow a webinar on the subject. Read more about this event here!