Locate hyperbolas on Ground Penetrating Radar slices with auto filter selection algorithm

Ground Penetrating Radar (GPR) is a geophysical method of using high frequency electromagnetic waves to study shallow objects below the ground. In the GPR sections, the position of point anomalies is represented as reflected hyperbola. In some cases, the reflected hyperbolas on the GPR cross-section have a relatively large number and overlap which making the positioning very complicated and time-consuming. This article presents the development of an automatic selection algorithm with the aim of accurately determining the position of hyperbolas on GPR cross-sections. The automatic selection algorithm is a combination of an edge detection algorithm and the establishment of filtering conditions based on image and physical properties of geological objects. The algorithm is applied to determine the hyperbolic position of several GPR cross-sections at specific locations and provides relatively accurate results. This result will enhance the accuracy in problems related to drawing tree root maps, underground construction, landslide surveys, subsidence, and soil pollution research.