Skip to main content
U.S. flag

An official website of the United States government

Official websites use .gov
A .gov website belongs to an official government organization in the United States.

Secure .gov websites use HTTPS
A lock ( ) or https:// means you’ve safely connected to the .gov website. Share sensitive information only on official, secure websites.

Unsupervised image segmentation for passive THz broadband images for concealed weapon detection

Published

Author(s)

Mabel Ramirez, Erich N. Grossman, Charles Dietlein, Zoya Popovic

Abstract

This work presents the application of a basic unsupervised classification algorithm for the segmentation of indoor passive Terahertz images. The 30,000 pixel broadband images of a person with concealed weapons under clothing are taken at a range of 0.8-2m over a frequency range of 0.1-1.2THz using single-pixel row-based raster scanning. The spiral-antenna coupled 36x1x0.02υm Nb bridge cryogenic micro-bolometers are developed at NIST-Optoelectronics Division. The antenna is evaporated on a 250υm thick Si substrate with a 4 mm diameter hyper-hemispherical Si lens. The NETD of the microbolometer is 125mK at an integration time of 30 ms. The background temperature calibration is performed with a known 25 pixel source above 330 K, and a measured background fluctuation of 200-5000mK. Several weapons were concealed under different fabrics: cotton, polyester, windblocker jacket and thermal sweater. Measured temperature contrasts ranged from 0.5-1K for wrinkles in clothing to 5K for a zipper and 8K for the concealed weapon. In order to automate feature detection in the images, some image processing and pattern recognition techniques have been applied and the results are presented here. We show that even simple algorithms, that can potentially be performed in real time, are capable of differentiating between a metal and a dielectric object concealed under clothing. Additionally, we show that pre-processing can reveal low temperature contrast features, such as folds in clothing.
Volume
6549
Conference Dates
April 9-13, 2007
Conference Location
Orlando, FL, USA
Conference Title
SPIE - The International Society for Optical Engineering, Defense & Security Symposium

Keywords

broadband images, clustering, concealed weapon detection, image processing, segmentation, terahertz

Citation

Ramirez, M. , Grossman, E. , Dietlein, C. and Popovic, Z. (2007), Unsupervised image segmentation for passive THz broadband images for concealed weapon detection, SPIE - The International Society for Optical Engineering, Defense & Security Symposium, Orlando, FL, USA, [online], https://tsapps.nist.gov/publication/get_pdf.cfm?pub_id=32672 (Accessed May 21, 2024)

Issues

If you have any questions about this publication or are having problems accessing it, please contact reflib@nist.gov.

Created April 8, 2007, Updated October 12, 2021