HOPS: A novel descriptor for Buried Target Detection in GPR images based on structural symmetry

Abstract

Buried target detection(BTD) is one of the most common applications in Ground Penetrating Radar(GPR) images due to the non-destructive detection advantages of GPR system and it remains a challenging task in GPR image analysis. Because GPR images are easily affected by different signal systems and noises. In this paper, we propose a new feature descriptor for ground penetrating radar(GPR) target detection named the histogram of orientated phase symmetry(HOPS), which is based on the symmetry and phase properties of GPR images. The proposed method was trained and evaluated using collected GPR data, these data consist of a variety of shapes, different metal content, and burial depths. We provide quantitative analysis that compares the receiver operating characteristics (ROCs) curve calculated by the proposed method with those calculated by the popular algorithms in BTD task at present, our algorithm has improved greatly.