An application of a particle filter to Bayesian multiple sound source tracking with audio and video information fusion
Hideki Asoh, Futoshi Asano, Takashi Yoshimura, Yoichi Motomura, Naoyuki Ichimura, Isao Hara, Jun Ogata, Kiyoshi Yamamoto · 2004
A particle filter is applied to the problem of detecting and tracking multiple sound sources by Bayesian inference using combined audio and video information. The problem is formulated within a general framework of Bayesian hidden variable sequence estimation by fusing observed information. The particle filter is then introduced as an approximation of Bayesian inference. Experiments using real-world data demonstrate that the proposed method works well in ordinary environments such as a meeting room. The computational cost of estimation is reduced significantly compared to exact Bayesian inference, while maintaining the quality of estimation.