Automatic corrosion classification and quantification of steel reinforcing bars within concrete using image data generated by an inductive sensor

This paper presents a methodology to automatically distinguish and quantify the corrosion of reinforcing bars within concrete using images generated by an inductive sensor. The methodology comprises three stages; image generation using the inductive sensor, image segmentation and feature extraction and neural network object classification. Preliminary results have shown that the methodology has correctly classified all the corroded parts on the testing samples while estimated the corrosion rate correctly on 80% of the testing sample