Share Email Print
cover

Proceedings Paper

A novel approach to automatic threat detection in MMW imagery of people scanned in portals
Format Member Price Non-Member Price
PDF $14.40 $18.00
cover GOOD NEWS! Your organization subscribes to the SPIE Digital Library. You may be able to download this paper for free. Check Access

Paper Abstract

We have developed a novel approach to performing automatic detection of concealed threat objects in passive MMW imagery of people scanned in a portal setting. It is applicable to the significant class of imaging scanners that use the protocol of having the subject rotate in front of the camera in order to image them from several closely spaced directions. Customary methods of dealing with MMW sequences rely on the analysis of the spatial images in a frame-by-frame manner, with information extracted from separate frames combined by some subsequent technique of data association and tracking over time. We contend that the pooling of information over time in traditional methods is not as direct as can be and potentially less efficient in distinguishing threats from clutter. We have formulated a more direct approach to extracting information about the scene as it evolves over time. We propose an atypical spatio-temporal arrangement of the MMW image data - to which we give the descriptive name Row Evolution Image (REI) sequence. This representation exploits the singular aspect of having the subject rotate in front of the camera. We point out which features in REIs are most relevant to detecting threats, and describe the algorithms we have developed to extract them. We demonstrate results of successful automatic detection of threats, including ones whose faint image contrast renders their disambiguation from clutter very challenging. We highlight the ease afforded by the REI approach in permitting specialization of the detection algorithms to different parts of the subject body. Finally, we describe the execution efficiency advantages of our approach, given its natural fit to parallel processing. mage

Paper Details

Date Published: 18 April 2008
PDF: 10 pages
Proc. SPIE 6948, Passive Millimeter-Wave Imaging Technology XI, 69480F (18 April 2008); doi: 10.1117/12.776943
Show Author Affiliations
Nitin M. Vaidya, Millivision Technologies (United States)
Thomas Williams, Millivision Technologies (United States)


Published in SPIE Proceedings Vol. 6948:
Passive Millimeter-Wave Imaging Technology XI
Roger Appleby; David A. Wikner, Editor(s)

© SPIE. Terms of Use
Back to Top