Share Email Print

Proceedings Paper

Low-power impulse signal classifier using the Haar wavelet transform
Author(s): James F. Scholl; Jonathan R. Agre; Loren P. Clare; Martin C. Gill
Format Member Price Non-Member Price
PDF $14.40 $18.00

Paper Abstract

Detection and classification of footsteps and other impulsive signals are a critical function of urban surveillance systems. An example is the wireless integrated network sensor (WINS) system, which is designed to meet this requirement for both law enforcement and military agencies. The detection and classification algorithms should be sufficiently robust to handle a wide variety of environments, but remain of low complexity to allow low power implementation. We present a modified time-domain method for impulse signal classification based on the Haar wavelet transform. The Haar wavelet basis is ideal for short time signals as it provides the best localization in the time domain. Further, the Haar transform has the shortest and simplest filter/basis system, with the scaling function filter using the average of two points and the wavelet filter being the difference between two points. Our classification scheme uses the Haar transform of the input signal to obtain the signal envelope, which is described by the decimated low pass filter coefficients. When implemented on many WINS nodes, this simple procedure has the further advantage of being able to do signal source detection in both location and time.

Paper Details

Date Published: 7 January 1999
PDF: 10 pages
Proc. SPIE 3577, Sensors, C3I, Information, and Training Technologies for Law Enforcement, (7 January 1999); doi: 10.1117/12.336958
Show Author Affiliations
James F. Scholl, Rockwell Science Ctr. (United States)
Jonathan R. Agre, Rockwell Science Ctr. (United States)
Loren P. Clare, Rockwell Science Ctr. (United States)
Martin C. Gill, Defence Science and Technology Organisation (Australia)

Published in SPIE Proceedings Vol. 3577:
Sensors, C3I, Information, and Training Technologies for Law Enforcement
Edward M. Carapezza; David B. Law, Editor(s)

© SPIE. Terms of Use
Back to Top