Novel means of salt concentration prediction and mapping on metallic surfaces using shortwave-infrared hyperspectral imaging

This preliminary work shows the potential application of hyperspectral imaging (HSI) to quantify salt concentrations on a metallic surface. Salt buildup on metals can accelerate corrosion processes on the surface - in particular, pitting corrosion. Higher concentrations of salt are linked to higher corrosion rates. In this work, magnesium chloride (MgCl2) salt solution is applied to aluminium coupons in concentrations ranging from 0 - 77.7 μg/cm2. These coupons are imaged using a hyperspectral sensor in the shortwave-infrared (SWIR) range [900 - 2500 nm] in 272 spectral bands. The actual salt concentration on the surface of the aluminium coupons is measured using a soluble salt tester (SST). Spectral angle mapper (SAM) values are calculated for each pixel in the image in relation to a reference MgCl2 spectrum collected using the same system. By using a novel thresholding algorithm based on these SAM values, we are able to predict the salt concentration on the surface with our model achieving an R2 value of 0.905. We are also able to visualise the spatial distribution of the MgCl2 in a way that outperforms an RGB imaging system and is not possible with the SST. This work is beneficial as it provides a non-contact solution for salt concentration measurement and offers the potential for many measurements across a large surface, reducing the risk of overlooking an area of high salt concentration which can be associated with localised surface measurements.