A novel approach for thermal crack detection and quantification in structural concrete using ripplet transform

Summary Fire accidents in the concrete structures affect the safety of human beings and structural components. Due to higher temperatures, thermal cracks are conceived in concrete elements. Crack identification, localization, and quantification are the three major components in the damage assessment system. This paper aims to detect thermal cracks of fire‐affected concrete structures using ripplet transform‐based computer vision method. Initially, the concrete images are decomposed with discrete ripplet transform (DRT). The low frequency sub‐bands are approximated with column filtering‐based gray level difference metric. It removes the uneven concrete background. Ripplet coefficient variance parameter (RCVP) is used to differentiate the crack pixels from noisy pixels which is used to eliminate the noises. Finally, in the reconstructed image, the major and minor thermal cracks are detected. The obtained thermal cracks are quantified with the crack properties of length, width, area, and perimeter. The novelty of the proposed method relays on the usage of ripplet transform for the detection of thermal cracks for different concrete grades and different durations of temperature profile. The experimental results are compared with four transform domain state‐of‐the‐method crack detection methods. The proposed method yields better results in terms of the accuracy and average execution time.