Underground object detection based on cross correlation and Hough transform

Abstract

GPR object detection is always a challenge in most practical applications. The hyperbola feature in B-Scan data is a very important feature for GPR object detection. In this paper, a novel method based on cross correlation between adjacent A-scans is proposed to detect reflection hyperbola. The method includes four steps. B-scan data is firstly preprocessed to suppress clutters. Then, the reflection hyperbola is extracted based on cross correlation. An approach of feedback is presented to fit the extracted hyperbola. At last, Hough transform is adopted to determine whether there is an object or not. The proposed method is tested by both simulated and field data.