3D Template Based Automatic Landmine Detection from GPR Data

摘要

In this paper, we present a 3D template based landmine automatic detection from GPR data. A 3D template is chosen and a 3D fuzzy template is designed. The choice of the 3D template is decided based on smooth changing position of the maximum amplitude at every C-scan as well as the threshold of its background average intensity. The 3D fuzzy template is extracted from 3D template crisp data. GPR similarity for both the 3D template as well as its fuzzy template is examined by crisp similarity measure and fuzzy similarity measure respectively. The cross correlation is applied as a similarity measure in crisp case while a fuzzy similarity measure, based on similarity of a measured data to a fuzzy rule, is applied in the fuzzy template case. Results of similarity applying both methods for automatic landmine detection are presented. The results show the promise of 3D fuzzy template and the fuzzy similarity measure in differentiating a landmine from other objects.