Login / Signup

Identification of Corrosion Minerals Using Shortwave Infrared Hyperspectral Imaging.

Thomas De KerfGeorgios PipintakosZohreh ZahiriSteve VanlanduitPaul Scheunders
Published in: Sensors (Basel, Switzerland) (2022)
In this study, we propose a new method to identify corrosion minerals in carbon steel using hyperspectral imaging (HSI) in the shortwave infrared range (900-1700 nm). Seven samples were artificially corroded using a neutral salt spray test and examined using a hyperspectral camera. A normalized cross-correlation algorithm is used to identify four different corrosion minerals (goethite, magnetite, lepidocrocite and hematite), using reference spectra. A Fourier Transform Infrared spectrometer (FTIR) analysis of the scraped corrosion powders was used as a ground truth to validate the results obtained by the hyperspectral camera. This comparison shows that the HSI technique effectively detects the dominant mineral present in the samples. In addition, HSI can also accurately predict the changes in mineral composition that occur over time.
Keyphrases
  • high resolution
  • machine learning
  • convolutional neural network
  • deep learning
  • photodynamic therapy
  • mass spectrometry