A deep learning framework based on improved self-supervised learning for ground-penetrating radar tunnel lining inspection

It is not practical to obtain a large number of labeled data to train a supervised learning network in tunnel lining nondestructive testing with ground-penetrating radar (GPR). To decrease the dependence of supervised learning on the number of labeled data, an improved self-supervised learning algorithm—self-attention dense contrastive...

Marchenko inversion of gpr data for a 1d dissipative medium

Radar data collected on two sides of a horizontally dissipative layered medium are required to invert for the medium parameters. The two-sided reflection and transmission responses are reduced to two single-sided reflection responses. One is the measured dissipative medium response, and the other is the reflection response of the...

High-resolution imaging and inversion of 3D wavefield data for layered media

Theory for 1D GPR data inversion for a dissipative layered medium

We present a data driven method of full waveform inversion in one dimension. This means that the inversion is carried out as a sequence of processing steps. The first step is known as Marchenko redatuming. In this step we retrieve focusing functions from the measured data. In the second step we isolate the last event in the focusing function to...