Image-Based Corrosion Detection in Ancillary Structures

Ancillary structures are essential for highways’ safe operationality but mainly prone to environmental corrosion. The traditional way of inspecting ancillary is manned inspection, which laborious, time-consuming, and unsafe inspectors. In this paper, a novel image processing technique was developed autonomous corrosion detection in-service structures. authors successfully leveraged features in the YCbCr color space as an alternative conventional red–green–blue (RGB) space. proposed method included preprocessing operation including contrast adjustment, histogram equalization, adaptive optimum value determination brightness. effect evaluated against semantically segmented ground truth set pixel-level annotated images. false rate higher Otsu than global threshold method; therefore, preprocessed images were converted binary using value. Finally, average accuracy true positive 90% 70%, respectively, achieved prediction