Explosive event discrimination
First Claim
Patent Images
1. A computer implemented method for determining whether an explosive event is the result of a launch or an impact, which method utilizes only a single acoustic sensor, the method comprising the steps of:
- collecting acoustic data resulting from the explosive event;
normalizing the acoustic data, which provides normalized acoustic signature data;
apply the normalized acoustic signature data to a discrete wavelet transform using a multiresolutional analysis, which provides detailed coefficients from multiple j level decompositions related to a blast wave, and approximation coefficients from multiple k level decompositions;
deriving a spectrogram of energy distribution from the approximation coefficients;
deriving a q-tuple feature vector space from the detailed coefficients;
deriving a m-tuple feature vector space from the spectrogram;
deriving a n-tuple feature vector space of energy distribution across normalized acoustic signature data;
combining the q-tuple, the m-tuple, and the n-tuple feature vector spaces to form a p-tuple feature vector subspace where p=q+n+m;
applying the p-tuple vector subspace to a trained classifier, to determine whether the explosive event was a launch event or an impact event; and
outputting an indication of the determination.
1 Assignment
0 Petitions
Accused Products
Abstract
A computer implemented method for discriminating between launch and impact events using acoustic sensors.
17 Citations
1 Claim
-
1. A computer implemented method for determining whether an explosive event is the result of a launch or an impact, which method utilizes only a single acoustic sensor, the method comprising the steps of:
-
collecting acoustic data resulting from the explosive event; normalizing the acoustic data, which provides normalized acoustic signature data; apply the normalized acoustic signature data to a discrete wavelet transform using a multiresolutional analysis, which provides detailed coefficients from multiple j level decompositions related to a blast wave, and approximation coefficients from multiple k level decompositions; deriving a spectrogram of energy distribution from the approximation coefficients; deriving a q-tuple feature vector space from the detailed coefficients; deriving a m-tuple feature vector space from the spectrogram; deriving a n-tuple feature vector space of energy distribution across normalized acoustic signature data; combining the q-tuple, the m-tuple, and the n-tuple feature vector spaces to form a p-tuple feature vector subspace where p=q+n+m; applying the p-tuple vector subspace to a trained classifier, to determine whether the explosive event was a launch event or an impact event; and outputting an indication of the determination.
-
Specification