Prediction method of underground pipeline based on hyperbolic asymptote of GPR image

Abstract

Ground penetrating radar (GPR) technology is one of the main techniques of the underground pipeline detection. We can analyze the information of the underground pipeline effectively and accurately by the ground penetrating radar data. There used to be three kinds of methods to predict pipeline position and diameter by hyperbola: using the method of least square curve fitting, using Hough transform to predict in parameter space, and using template matching method. This paper will begin with a model, and analyze several factors affecting the pipeline detection. According to the model and the factors, we put forward a new method to calculate the position and diameter of pipeline. The difference between this method and the traditional method is to calculate with the help of the hyperbolic asymptote, and to reduce factitious factors by a clustering method. The image of underground target from the movement of GPR will generate edge information like hyperbola. We can extract the hyperbola from the image by extreme value methods or differential operator methods. In contrast, our method neither needs to pre design templates, nor needs complex calculation. It is suitable for real-time data processing of GPR.