| Popis: |
In this paper a method for hyperbolic target detection in subsurface radar images is presented. The method is based on a suitable fuzzy clustering approach aimed to individuate partially masked and/or corrupted target signatures by exploiting a fuzzy-shell algorithm. Fuzzy logic allows to exploit as much as possible the information hidden in the data, while the robust clustering approach provides a way not only to detect, but also to validate the results obtained. The procedure is applied, with an overall correct classification rate of 84%, to actual GPR data at 400 MHz. |
