Reactive Intensity Vector Based Direct Path Detection for DoA Estimation on a Single Acoustic Vector Sensor
Bing Zhu, Wen Zhang · 2023
As a compact device, an acoustic vector sensor (AVS) has been widely used for source direction of arrival (DOA) estimation in 3D space. However, when deployed in complex environments, such as with strong noise, reverberation, and multiple sources, the DOA estimation accuracy degrades significantly. In this paper, we propose an efficient method for multi-source DOA estimation based on an AVS in a reverberant room environment. The method is based on exploiting a feature from the AVS signals, i.e., the reactive intensity vector, to detect the direct-path dominated parts of the received signals in time-frequency bins. Then, intensity-vector based DOA estimation is performed, which has low computational complexity. As demonstrated through experiments, the proposed method outperforms the state of the art methods even in highly reverberant environments.