Using radar in a through-the-wall imaging application is an expanding field of research both for civilian and military uses. Thus far, most of the attention has been directed toward building radar imaging systems to detect objects within a room or building. The resulting images are full of ambiguity and difficult to interpret what the image is displaying. Presented here is a novel approach that addresses the interpretation of the images produced by the aforementioned imaging systems. We propose a classification scheme that provides an interpretation of an urban environment imaged in 3D. This approach builds probabilistic object models from feature vectors extracted from a volumetric radar image. A minimum-distance classifier is used to label radar image data and provide a 3D visualization of an urban scene. Results using real radar backscatter data validate the effectiveness of our method.

Date Published: 16 April 2008
PDF: 10 pages
Proc. SPIE 6943, Sensors, and Command, Control, Communications, and Intelligence (C3I) Technologies for Homeland Security and Homeland Defense VII, 69430T (16 April 2008); doi: 10.1117/12.777883

Published in SPIE Proceedings Vol. 6943:
Sensors, and Command, Control, Communications, and Intelligence (C3I) Technologies for Homeland Security and Homeland Defense VII
Edward M. Carapezza, Editor(s)