Researchers have developed a new drone-compatible method that can detect corrosion hiding beneath painted metal surfaces without needing to heat the material.
The technology could make inspections of ships, coastal infrastructure and industrial facilities faster, safer and less expensive.
Iron and steel are widely used to build ships, bridges, ports and other large structures because they are strong and durable.
However, exposure to moisture, salt and other environmental conditions can gradually cause these metals to corrode.
If corrosion continues unnoticed, it can weaken a structure and eventually contribute to serious failures.
Regular inspections are therefore essential, particularly in maritime environments where saltwater can accelerate the corrosion process.
Traditional visual inspections have limitations. They can take considerable time, especially when workers need to examine large ships or difficult-to-reach structures.
More importantly, corrosion may begin underneath layers of paint or protective coatings, making it impossible to detect simply by looking at the surface.
Drones equipped with cameras offer one possible solution. They can quickly inspect large areas and reach locations that may be difficult or dangerous for people. However, existing drone-based technologies also have drawbacks.
Multispectral cameras, which capture information from different wavelengths of light, have generally been used to identify corrosion that is already visible on a surface.
Thermal cameras can detect some corrosion hidden beneath paint, but these systems typically require the surface to be heated first. That makes inspections more complicated and limits where the technology can be used.
In a new study published in PNAS Nexus, Emma Hernández-Suárez and her colleagues developed a different approach that can identify both visible and hidden corrosion without heating the metal.
The researchers discovered that information collected from only three specific wavelengths of light—518, 655 and 838 nanometers—could provide enough information to detect corrosion when combined with a measurement of surface texture.
They used this information to develop an algorithm capable of distinguishing healthy metal from areas affected by corrosion.
To train the system, the researchers created corrosion samples inside a salt spray chamber, which reproduces the harsh conditions that can cause metals to deteriorate. They then tested the algorithm using a separate collection of samples taken from an industrial environment. Importantly, these samples had different types and colors of protective coatings that were not known to the algorithm beforehand.
The system achieved more than 90% accuracy in detecting both surface corrosion and corrosion hidden beneath coatings. It continued to perform well across different coating materials and colors, suggesting that the technique could work in a wide variety of real-world situations.
The researchers say the algorithm is freely available and can work with commercially available multispectral camera systems that can be mounted on drones.
This could allow ships, ports and other large steel structures to be inspected more efficiently while reducing the need for workers to enter dangerous or difficult-to-access areas. By identifying hidden corrosion before serious damage develops, the technology could also lower maintenance costs and help prevent structural failures.
