Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
383731 | Expert Systems with Applications | 2014 | 8 Pages |
•Pattern recognition approach to shock wave and muzzle blast classification.•Two-stage shock wave and muzzle blast recognition.•Independent from ballistic model, wavelet based signature representation.•Approximation coefficients thresholding for classification purposes.
The article presents a pattern recognition approach to acoustic shock wave and muzzle blast detection. Gunshot signatures are divided into multiple classes, given by combination of 3 types of supersonic weapons of different caliber: 7.62 mm, 5.56 mm and 9 mm and 3 types of acoustic events: shock wave, muzzle blast and reflections. The classification is performed on wavelet compressed 100 μs time frames. The experiment shows that the choice of a fitting wavelet base is crucial for the quality of recognition.