Symmetry based 3D GPR feature enhancement and extraction

The efficient analysis of high-resolution 3D GPR data sets is of increasing importance given todays possibilities to acquire large and dense data volumes. In order to reduce data complexity and enhance the features of interest, attribute based analysis, a well-established field in reflection seismology, has received growing interest also in the GPR community. Here, we present a novel GPR attribute called phase symmetry, which is adapted from image processing. We believe that this attribute is well-suited for analyzing 3D GPR data sets. In this study, we introduce the basic concepts of phase symmetry. Using two synthetic examples and comparing phase symmetry to the well- known Canny edge detector, we illustrate that phase symmetry also provides high quality results in the presence of significant noise and smoothly varying anomalies. Using two 3D GPR field examples collected to detect buried utilities and archaeologically relevant features, respectively, demonstrates the applicability of phase symmetry to real GPR data and illustrates that this attribute is an effective tool in order to extract symmetric features embedded within a heterogeneous background. Additionally, we show that phase symmetry can be combined with the similarity attribute to jointly emphasize event symmetry and waveform similarity.