Robust Acoustic Localization Via Time-Delay Compensation and Interaural Matching Filter
Jie Zhang,Hong Liu
DOI: https://doi.org/10.1109/tsp.2015.2447496
IF: 4.875
2015-01-01
IEEE Transactions on Signal Processing
Abstract:Acoustic localization is an essential technique in speech capturing, speech enhancement, video conferencing, and human-robot interaction. However, in practical situations, localization has to be performed in abominable environments, where the presence of reverberation and noise degrades the performance of available position estimates. Besides, the designed systems should be adaptive to locomotion of targets with low computational complexity. In the context, this paper introduces a robust hierarchical acoustic localization method via time-delay compensation (TDC) and interaural matching filter (IMF). Firstly, interaural time-delay (ITD) and interaural level difference (ILD), which are cues involved in first two layers, respectively, are yielded by TDC all at once. Then, a novel feature named IMF, which can eliminate the difference between binaural signals, is proposed in the third layer. The final decision making is based on a Bayesian rule. The relationships among the three layers are that the former layer provides candidate directions for later ones such that the searching space becomes gradually smaller to reduce matching time. Experiments using both a public database and a real scenario verify that TDC and IMF are robust for acoustic localization, and hierarchical system has less consumption time.