Infrared thermography model for automated detection of delamination in RC bridge decks

Abstract

The analysis of Infrared thermography (IRT) data to evaluate the condition of concrete bridge decks still relies on user interpretation of thermal images. Delaminations are commonly defined based on the pixels associated with temperatures higher than predefined threshold values that are arbitrarily selected. Such a subjective scheme can produce inconsistent results. The present study proposes a robust automated procedure for detecting and classifying delaminations in concrete bridge decks. IRT testing was conducted in-situ on full-scale bridge decks. The thermal images were enhanced and a stitching algorithm was developed using especially written codes to create a mosaicked thermogram of the entire bridge deck from individual images. Image analysis based on the k-means clustering technique was utilized to segment the mosaic and identify objective thresholds. Hence, a condition map delineating different categories of delamination was created. The proposed procedure was implemented in four full-scale bridge decks and validated via the results of other technologies obtained on the same bridges. The proposed approach resulted in effective identification of delaminated areas within the bridge decks. The findings could stimulate wider acceptance of IRT as a rapid, systematic and cost-effective evaluation technique for critical bridge deck deficiencies, allowing to focus maintenance and repair budgets on most deserving bridge decks.