Performance assessment of generalized cross-correlation based algorithms for multisource point-based localization and detection

In this study, we develop two new algorithms based on generalized cross-correlation (GCC) approach for improving multisource detection-localization performance. The algorithms are based on time-delay estimation using classical cross-correlation (CC) and smoothed coherence transform (SCOT) methods. Beside assessment of the two proposed GCC methods with the existing GCC-PHAT method, peformance of the FAST SRP-PHAT method is also compared with group of three GCC-based methods in terms of localization precision and computing cost. The evaluation results show that: (i) The method based on CC algorithm outperforms two other methods based on SCOT and GCC-PHAT algorithms while peformance of the SCOT method is quite similar to the GCC-PHAT method. (ii) Each computing cost of the three GCC methods is less than the computing cost of FAST SRP-PHAT method but the GCC methods' localization precision is degraded.