Drone-compatible method detects corrosion under paint without heat

Corrosion poses a significant threat to vital infrastructure such as merchant ships and coastal facilities, which are often constructed from iron and steel. While visual inspections are standard, they can be laborious and imprecise. Drones equipped with multispectral cameras offer a promising alternative, but existing methods primarily detect surface corrosion or require heating to identify subsurface damage.

Researchers have developed a novel approach that overcomes these limitations. By analyzing information from just three specific spectral bands (518 nm, 655 nm, and 838 nm) in conjunction with a texture feature, the new algorithm can accurately detect both surface and subsurface corrosion without the need for sample heating. This method is particularly beneficial for painted structures where corrosion can develop unseen.

The corrosion-detection algorithm was trained on data from samples artificially corroded in a salt spray chamber. It was then rigorously tested on a separate set of samples from an industrial setting, featuring various unknown coatings. The algorithm demonstrated consistent performance, achieving an accuracy rate exceeding 90% across different types and colors of coatings.

This freely available algorithm is designed for compatibility with commercial multispectral systems that can be mounted on drones. The developers suggest that its implementation can lead to reduced inspection costs and enhanced safety for maritime and industrial assets by providing a more efficient and effective means of monitoring structural integrity.