Comparison of thermal and hyperspectral data to correlate heat maps with spectral profiles from galvanized steel surfaces

A hyperspectral and thermal imaging system have been used to acquire hyperspectral data and thermal images from the surface of galvanized steel parts. After heating the parts in an oven, heat maps have been partitioned into hot (red) and cold (blue) regions as relative representation of heat distribution, to correlate the spectral signatures from hyperspectral data to the hot and cold areas. Three spectral point ratios were extracted as effective features from each spectrum and used for classification. We present the classification results for spectra representing irregularities in the galvanization process. Preliminary results suggest that hyperspectral imaging can potentially be used for surface characterization of galvanized metal surfaces as a nondestructive testing or remote inspection tool.