Buried pipe localization using an iterative geometric clustering on GPR data

摘要: Ground penetrating radar is a non-destructive method to scan the shallow subsurface for detecting buried objects like pipes, cables, ducts and sewers. Such cause hyperbola shaped reflections in radargram images achieved by GPR. Originally, those were interpreted manually human experts an expensive time consuming process. For acceleration of this process automatization interpretation desirable. In paper efficient approach recognition pipe localization radargrams presented. The core our iterative directed shape-based clustering algorithm combined with sweep line using geometrical background knowledge. Different recent state art methods, able ignore noise recognize multiple intersecting or nearby hyperbolas without prior knowledge about number pipes. whole deliver position estimates error only few millimeters, as shown experiments two different data sets.