Distorted Hyperbola Extraction of GPR Data using Edge Detection for Buried Non-Metallic Pipes

Non-metallic pipes are widely used in the urban underground pipe network. To map the pipeline system, obtaining the accurate locations and directions of the pipelines is an essential challenge. In this work, Ground Penetrating Radar (GPR) is used to detect buried non-metallic pipes since it is an efficient and non-destructive method for buried non-metallic pipes. The main objective of this paper is to develop a simple, efficient and automated method to locate non-metallic pipelines, by considering the estimation of the buried depth and the distortion of hyperbola shown on the B-scan. The radargrams of buried non-metallic pipes with several angles (0°, 30°, 60° and 90°) between the GPR survey line and the pipeline were compared numerically and experimentally. Edge detection with Canny operator was used for the automatic extraction of the hyperbolas reflected by buried pipes. The pixels of the edge detection image were adapted to the detection area. So, the buried depth of the pipes could be estimated by picking the pixel point of the hyperbola vertex accurately. The estimation results showed good accuracy with good shaped and distorted hyperbolas. It can also be extended to the detection of other buried objects.