Corrosion Severity Index (CSI) for Spectral Characterization of Corroded Steel and Iron Samples

Abstract

Metals in coastal regions are threatened by corrosion, underscoring the need for precise detection and monitoring methods. Traditional methods often face limitations in terms of accuracy and applicability under diverse conditions. This study introduces the corrosion severity index (CSI), an innovative spectral index for assessing corrosion in steel and iron structures. Several iron samples were placed in a salt spray chamber to generate different degrees of corrosion. The samples were analyzed using hyperspectral cameras covering the visible near-infrared (VNIR) to the shortwave infrared (SWIR) spectrum. A scale-invariant feature transform (SIFT) registration algorithm was employed to generate the full spectral signatures from 400 to 1700 nm for each pixel. The CSI combines four spectral bands (457.50, 791.91, 1305.08, and 1442.60 nm) where a pixel value close to 0 represents the absence of corrosion, whereas a higher value indicates greater severity of corrosion. Based on the average CSI values, samples are classified into Grade A, B, C, or D, which indicates the degree of corrosion. CSI demonstrates its ability to detect early-stage corrosion and has been evaluated for robustness across a variety of steel and iron samples in different environmental conditions. In addition, the performance of the CSI is validated by comparing it with the previously published corrosion index (CI). CSI demonstrates a higher accurate ability to detect corrosion products and identify the degree of corrosion with a simplified approach. This index allows a balance between accuracy, low computational demands, and usability, providing an optimal solution for early diagnosis and proactive management of corrosion in coastal infrastructures.