Single-Source-Point Detection for DOA Estimation Using Angle Correlation Between Adjacent Time–Frequency Points

Abstract

This letter proposes multisource direction-of-arrival (DOA) estimation using the correlation between angles of adjacent time– frequency (TF) points for a first-order ambisonics sensor array. For a TF point in the recorded signal, we define the adjacent TF points whose angles are close to that of this point as angle correlation points (ACPs) and then explore the relation between the probability that this point is a single-source point (SSP) and the number of ACPs. We found that there is a positive correlation between the number of ACPs and the probability that a point is an SSP. Hence, SSP detection is proposed using the angle correlation between adjacent TF points. In addition, 2-D weight kernel density estimation is designed to estimate the probability density of angles of detected SSPs. Finally, peak search is adopted for DOA estimation. Experiments in simulated and real environments show that the DOA estimation performance of the proposed method exceeds that of some existing methods.