Restoration of the distorted color to detect the discoloration status of a steel bridge coating using digital image measurements

Discoloration, representing the degradation of the coating, is often a trigger of a progressive damage process for a steel bridge. The detection of discoloration status often relies on a routine visual inspection in Taiwan, resulting in subjective results. To minimize human error in such inspection, this study provides an alternative approach, in which the digital image measurement is adopted. The process of acquiring images in any means is, in fact, a duplicate of the cross-media color process and often results in distorted colors. Consequently, to determine the discoloration between fading and its authentic color, one first must perform the color restoration for the acquired faded images. To reduce error in the restoration process, this study uses the least-square support vector machine (LS-SVM), spectral power distribution, spectral reflectance and matrix restoration to form an integrated algorithm. The average color difference between distorted and undistorted fading color of the proposed approach is approximately 5 (NBS value), which is applicable in the industry, requiring a 4th level of color difference for steel bridge coating inspections. A smartphone application is developed based on the algorithm established to facilitate the application for detecting color differences in steel bridges.