@article {Wu:2018:0736-2935:1117,
title = "Reconstruction of Radiated Noise Demodulation Spectrum by Exploiting the Structure of Group Sparsity",
journal = "INTER-NOISE and NOISE-CON Congress and Conference Proceedings",
parent_itemid = "infobike://ince/incecp",
publishercode ="ince",
year = "2018",
volume = "258",
number = "6",
publication date ="2018-12-18T00:00:00",
pages = "1117-1126",
itemtype = "ARTICLE",
issn = "0736-2935",
url = "https://ince.publisher.ingentaconnect.com/content/ince/incecp/2018/00000258/00000006/art00014",
author = "Wu, Qisong and Xu, Ping and Fang, Shiliang",
abstract = "It is well known that there exists obvious amplitude modulation on the wideband radiated noise due to the propeller rotation; therefore, the feature of demodulation spectrum is of great importance in underwater acoustic target classification or recognition application. In this paper,
a novel reconstruction method of high-resolution demodulation spectrum is proposed by exploiting the underlying structure of group sparsity. On one hand, the estimation of demodulation spectrum is converted into the reconstruction of sparse coefficient vector by taking the sparse characteristic
of demodulation spectrum into consideration; on the other hand, the group-sparsity structure of demodulation spectra in the multiple subbands is exploited to improve the reconstruction performance. Compared to the FFT based demodulation spectrum, in which the sequential operations, including
line spectra detection and feature extraction, are inevitable carried out for recognition or classification, the reconstructed sparse demodulation spectrum automatically provides the feature of line spectra such as, frequency and amplitude. It avoids the annoying threshold parameter tuning
in the detection and manual intervention in the feature extraction. Furthermore, the resolution of the reconstructed demodulation spectrum is obviously superior to that in the FFT based demodulation spectrum which is confined to the Rayleigh bound in the Bayesian sparse reconstruction framework.",
}