Detecting Pinholes in Coatings with Hyperspectral Imaging

Coatings are critical to maintaining the integrity of industrial and offshore equipment, but defects reduce their effectiveness. Pinholes, locations on a coated surface which remain uncoated, are one kind of defect. They difficult to detect because of their small size. Here we investigate the possibility of using Hyperspectral imaging to find pinholes. Seven painted steel panels are imaged with a short wave infrared hyperspectral camera. Two target detection algorithms are tested for finding pinholes using two spectral signatures: steel and an underlayer of epoxy paint. One pinhole that occupies multiple pixels is identified, and several possible subpixel pinholes are detected for each target signature. We expect that further improvements to the characterization of the target signal, a better understanding of false positives and noise and improvements to the validation process will allow the detection of subpixel pinholes.