Due to the rapid construction of social infrastructure in the 20th century, automation of maintenance and management of social infrastructure is an important issue in the 21st century, as it is aging simultaneously. The use of ground-penetrating radar to understand and automatically analyze information in the underground space is attracting attention as an important tool in the maintenance and management of buried pipes, but analytical methods have not yet been developed. This study proposes an automatic detection method for underground pipes using hyperbolic fitting. Since the parameters can be set freely, this method is expected to detect buried pipes with higher accuracy than conventional artificial intelligence models, which are easily affected by the quality of training data. It is also shown that the detection rate can be improved by using three-dimensional filtering and edge detection, and that it is possible to observe temporal changes in the state of buried pipes.