Signal processing algorithms in identification of subsurface objects
O. V. Sytnik, A. V. Gorohovatsky · Radioelectronics and Communications Systems · 2007
This study presents algorithms of preliminary signal processing of subsurface probing radars and an identification technique of objects under observation. Preliminary processing consists of filtration, removal of unwanted nonstationary trends, segmentation and search for connected regions. The identification technique is based on the tools of the Hough class transform. Such approach is founded on the possibility of presenting electromagnetic signal reflected from physical object in the form of analytical model in the space of parameter spectra. Preliminary data processing is performed in the interactive mode ensuring the minimum signal-to-noise ratio when presenting objects in the space of characteristic functions. A procedure for analyzing radar images with respect to connectivity of their elements was proposed. Arguments regarding the choice of thresholds in the space of parameter spectra and the space of signals have been considered. An algorithm of image convolution was developed to enhance the contrast-background sensitivity.