An automatic algorithm for multi-defect classification inside tunnel using SVM

Abstract

An integrated framework is presented in this paper to automatically achieve rebar detection and defection classification inside tunnel. This framework is composed of GPR return preprocessing to perform clutter reduction, a Frequency-wavenumber migration algorithm to focus the hyperbola, an energy scanning method to extract the region of interest(ROI) and to achieve rebar detection, and a multi-class support vector machine(SVM)to classify various types of defection inside tunnel. The experimental results based on simulated data show that the presented framework can automatically and effectively perform rebar detection and defection classification.