Development of a change-detection method on underground air void identification by time-lapse GPR 3D imaging

This study reviews the parameters that affect GPR 3D imaging quality and proposed a guideline for establishing time-lapse GPR slice generation to detect air voids. A lab control experiment was conducted with mutual information applied to evaluating the reliability of proposed imaging workflow. Reflectance thresholding and structural similarity were explored in identifying air voids from time-lapse GPR slice series and got promising result. It was found that variation of GPR slice images come from ununiform reflection intensity. Image change detection method is feasible in extracting features from speckled GPR time lapse slices. And a more convincible GPR slice generation workflow contributes to more accurate temporal feature extraction from GPR time lapse slice.