Tracking of Multiple Sound Sources by using Mean Shift Algorithm for Arbitrary Microphone Array Configuration
Tracking of Multiple Sound Sources by using Mean Shift Algorithm for Arbitrary Microphone Array Configuration
カテゴリ: 部門大会
論文No: OS4-8
グループ名: 【C】平成25年電気学会電子・情報・システム部門大会講演論文集
発行日: 2013/09/04
タイトル(英語): Tracking of Multiple Sound Sources by using Mean Shift Algorithm for Arbitrary Microphone Array Configuration
著者名: Riyo Ishida(Keio University),Yasue Mitsukura(Keio University),Nozomu Hamada(Keio University)
著者名(英語): Riyo Ishida(Keio University),Yasue Mitsukura(Keio University),Nozomu Hamada(Keio University)
キーワード: 音源追跡|時間周波数解析|ミーンシフト|マイクロホンアレー|sound source tracking|time-frequency analysis|mean shift|microphone array
要約(日本語): In this paper, we proposed the novelty method of tracking multiple sound sources using the direction-of-arrival (DOA) estimation method based on the sparsity of sound signals in time-frequency domain and weighted mean shift algorithm. It is difficult to apply the conventional DOA estimation method to track the moving sound sources because moving sound sources should be estimated within a short time interval. Mean shift method based on a kernel density estimation is able to increase the reliability of DOA estimation by using only the observed data. The extended mean shift algorithm using weights for previous frames and pause detection are proposed to improve the tracking accuracy. In order to show the effectiveness of the proposed method, we demonstrate the simulation of tracking.
PDFファイルサイズ: 6,147 Kバイト
受取状況を読み込めませんでした
