Use of modulation spectra for representation and classification of acoustic transients from sniper fire

Abstract

There are many applications for classification of acoustic transients produced by supersonic projectile fire. Analysis of existing models for such transients suggests they have properties which may be well-captured by a transform of a signal into joint acoustic and modulation frequency: a modulation spectral representation. Simple features are extracted from this representation which enables successful use in an important classification application. 1. INTRODUCTION Automated or semi-automated identification and localization of snipers has obvious military and safety applications. Acoustic signatures from rifle shots are loud and distinctive, making them nominally easy to detect [1]. Passive acoustics sensing provides good range and non-line-of-sight capabilities [2]. Despite its many advantages, the performance of acoustic sensors can degrade under adverse propagation conditions. In particular, the direct weapon muzzle blast, which originates from the shooter location and the projectile shockwave, which does not relate to that location, overlap in frequency. Both signals are broadband and there is little, if any, reliable difference in their temporal and spectral patterns. Localization accuracy can be degraded by this overlap. In this paper we add another signal dimension, modulation frequency, to potentially allow for increased discrimination between the muzzle blast and the shockwave. We show that this added dimension significantly increases discrimination accuracy over spectral features alone. In the remainder of paper, we first describe and illustrate the nature of the muzzle blast transient. We then describe a system which jointly estimates standard acoustic frequencies along with modulation

DOI: 10.1109/ICASSP.2005.1416212

Extracted Key Phrases

6 Figures and Tables

Showing 1-3 of 3 references

Acoustic sniper localization system

  • Gervasio Prado, Hardave Dhaliwal, Philip O Martel

Optimal and wavelet-based shock wave detection and estimation