Application of subspace decomposition in feature extraction of acoustic targets

Aim To investigate the technique of feature extraction for helicopter identification. Methods Based on the analysis of spectral characteristics of a helicopter acoustic signal, a method of subspace decomposition was proposed for the purpose of suppressing noise interference and increasing the ability of recognizing target and multiple signal classification (MUSIC), an algorithm of harmonic retrieval, was introduced to extract harmonic frequencies from acoustic signals. Results and Conclusion By using MUSIC harmonic frequencies were extracted from really measured helicopter acoustic signals. The results show that it is feasible to extract features in frequency domain from acoustic signals with the method of subspace decomposition and MUSIC is an effective algorithm of extracting harmonic frequencies.