Detection of corrosion on steel structures using automated image processing
Mojtaba Khayatazad, L. De Pue, Wim De Waele
Abstract
The traditional method used for corrosion damage assessment is visual inspection which is time-consuming for vast areas, impossible for inaccessible areas and subjective for non-experts. A promising way to overcome the aforementioned drawbacks is to develop an artificial intelligence-based algorithm that can recognize corrosion damage in a series of photographic images. This paper reports on the implementation and use of an algorithm that quantifies and combines two visual aspects – roughness and color – in order to locate the corroded area in a given image. For the roughness analysis, the uniformity metric calculated from the gray-level co-occurrence matrix is considered. For the color analysis, the histogram of corrosion-representative colors extracted from a data-set in HSV color space is used. The algorithm has been applied to a large dataset of photographs of corroded and non-corroded components and structures. Our findings show that the developed algorithm can efficiently locate corroded areas.