Damage Quantification Through Digital Images of Cracked Concrete Surfaces Presenting Anisotropic Behaviour

Digitization of established quantitative visual inspection (i.e., cracking index - CI) protocols have been evaluated. Datasets to train machine learning models are limited, and even more so for quantitative purposes. Automation of the CI has revealed some primitive challenges at the data collection stage. Therefore, this work focused on an image acquisition and processing standard approach to ensure valuable quantitative data can be extracted from the images. The anisotropy of laboratory made concrete blocks was further captured through image analysis which sheds light on the need to quantify anisotropy to efficiently select coring locations/configurations that best captures damage through further testing. This work showed that digital images can provide CIs similar to those obtained directly from the concrete surface provided that a certain resolution is kept. Moreover, the anisotropic behaviour was observed through a divided CI where cracks preferentially follow the reinforcement and through crack orientations from image analysis.