Corrosion detection and evaluation for steel wires based on a multi-vision scanning system

Cable replacement has been commonly implemented to avoid its insufficient bearing capacity after long-term service. To guide better the cable maintenance and life assessment, the experimental studies on replaced cables are significant. As the fundamental step of replaced cable assessment, the corrosion examination of steel wires is required. The existing methods and systems mainly focus on the qualitative estimation and localization of corrosion, which is hard to fulfill the requirement of precise analyses. In this paper, a multi-vision scanning system is proposed to detect and represent the real corrosion and its distribution on the surface of a steel wire. Applied with a novel high-definition image scanning hardware and advanced panoramic image stitching processing algorithms, the surface 2D texture of a single steel wire can be scanned automatically. Afterward, by using the deep learning methods, the corrosion is precisely identified. Furthermore, non-uniformity indicators are proposed for the description of the apparent corrosion. The system has been verified and utilized in the inspection and evaluation of several steel wires from a replaced suspender. Due to it is effective and economical, the system is promising and serves as an instrument for both corrosion researches and structural assessment.

Automated summary

A multi-vision scanning system is proposed in this paper for detecting and representing the real corrosion and its distribution on the surface of a steel wire. By utilizing high-definition image scanning hardware and advanced image processing algorithms, the surface texture of a single steel wire can be automatically scanned and corrosion can be accurately identified using deep learning methods.