Machine vision-based model for spalling detection and quantification in subway networks

Spalling is a significant surface defect that can compromise the integrity and durability of concrete structures. The detection and evaluation of spalling are predominantly conducted on the basis of visual inspection (VI) techniques. Although these methods can provide substantial inspection data, they are known to be time-consuming, costly, and qualitative in nature. The objective of this paper is to develop an integrated model based on image processing techniques and machine learning to automate consistent spalling detection and numerical representation of distress in subway networks. The integrated model consists of a hybrid algorithm, interactive 3D presentation, and supported by regression analysis to predict spalling depth. First, RGB images are preprocessed by means of a hybrid algorithm to de-noise the image and enhance the crucial clues associated with spalling. Second, a spalling processor is designed to detect distress attributes, thereby providing 3D visualization model of the defect. And third, the depth and severity of spalling distress are measured using a novel regression analysis model in conjunction with image processing techniques in intensity curve projection. The integrated model was validated through 75 images. Regarding the hybrid algorithm, the recall, precision, and accuracy attained, were 91.7%, 94.8%, and 89.3% respectively. The mean, standard deviation of error percentage in spalling region extraction were 11% and 7.1% respectively, while the variance was 25. Also, the regression model was able to satisfactory quantify the spalling depth with an average validity of 93%. The integrated model is a decision support tool, expected to assist infrastructure managers and civil engineers in their future plans and decision making.