Acoustic Source Track Before Detect Algorithm Using Dynamic Programming
Yao Guo, Feng Lin, Hongyan Zhu · 2023
Aiming at the problem of tracking acoustic source with weak signal, dynamic programming (DP) is introduced to implement the Track Before Detect (TBD) task. In this method, a low threshold is used to obtain potential steered response power (SRP) measurements, and a recursive way is adopted to accumulate the source energy and to associate the source positions between adjacent frames. To reduce the miss detection, the predictive positions have also been treated as the measurements. Experiment results demonstrate that the proposed method can achieve more accurate position estimate than the existing TBD implemented by Particle Filter (PF) in moderate reverberation and noisy conditions.