Automatic optical & laser-based defect detection and classification in brick masonry walls

Abstract

A real time system fusing data from vision and laser sensors to detect and classify types of defects in brick masonry is presented. A Support Vector Machine (SVM) algorithm is conceived and used to develop a Defect Finding Classification Model (DFCM) to automatically classify the types of defects found in masonry walls using the image data obtained from both vision and 2D laser sensors mounted on an articulated 6-axis robotic arm. Thirteen image features were extracted to train the SVM. It was found that the proposed approach has a detection accuracy of over 96%.