Clutter reduction using image-based thresholding approach for 3-D through-wall synthetic aperture radar

PDF

Authors
  1. Sévigny, P.
Corporate Authors
Defence Research and Development Canada, Ottawa Research Centre, Ottawa ON (CAN)
Abstract
Concealed targets, such as humans within buildings, can potentially be detected using through-wall radar imaging technologies. The success of the human target detection task highly depends on the amount of clutter present in the imaged scene. The goal of this study is to significantly reduce the amount of clutter found in through-wall radar imagery, so as to automatically identify potential candidate targets for human target classification algorithms. To do so, all the connected components or blobs in the imagery are identified by their intensity and location. The first proposed thresholding method uses a global threshold set to the median of the blob maximum intensities. This thresholding method reduces the number of blobs by 50%. The second proposed thresholding method uses the blob locations as seeds for a region-growing algorithm. This automatically removes large and strong features that could otherwise overwhelm and compromise target classification algorithms. The thresholding methods were tested using experimental data collected from seven different buildings of various wall materials. They were found to be fast, robust, and capable of discarding large amounts of clutter while keeping human target signatures for all scenes except one. Previously, the candidate targets had to be selected and segmented manually by a radar specialist. The thresholding methods introduced in this Scientific Report thus represent a significant advance in through-wall radar image data exploita

Il y a un résumé en français ici.

Keywords
through-wall radar;image thresholding;clutter reduction
Report Number
DRDC-RDDC-2015-R278 — Scientific Report
Date of publication
01 Dec 2015
Number of Pages
86
DSTKIM No
CA041923
CANDIS No
803082
Format(s):
Electronic Document(PDF)

Permanent link

Document 1 of 1

Date modified: