Using Karhunen - Loeve filter in the processing of ground penetrating radar data (GPR)

Abstract

Ground - penetrating radar (GPR), a modern geophysical method, is used to study the shallow structure below the ground, based on the principle of electromagnetic wave generation and reception in the high frequency range of 10 - 3000 MHz. The received signal could be presented in the form of images, similar to the ultrasound method in medicine, to study the shallow geological structure below the ground. In order for the GPR cross - section to accurately reflect the environment below the ground, data processing always plays a very important role, in order to eliminate horizontal noise, permanent noise, etc., which often blur or obscure objects of interest. Many signal processing methods have been studied and applied, but the improvement of processing methods has always received much attention from scientists. This paper presented the theoretical basis and how to apply the Karhunen - Loeve filter to the model and practical GPR cross - section, in order to remove the standing and horizontal noises. Experimental results on the theoretical and practical models showed that the interested signals were highlighted and some noise forms were blurred. This initial success would be a promise to develop the problem of automatic processing and identification in the GPR method of the research team.