Acoustic Vector Sensor Multi-Source Detection Based on Multimodal Fusion

Abstract

The direction of arrival (DOA) and number of sound sources is usually estimated by short-time Fourier transform and the conjugate cross-spectrum. However, the ability of a single AVS to distinguish between multiple sources will decrease as the number of sources increases. To solve this problem, this paper presents a multimodal fusion method based on a single acoustic vector sensor (AVS). First, the output of the AVS is decomposed into multiple modes by intrinsic time-scale decomposition (ITD). The number of sources in each mode decreases after decomposition. Then, the DOAs and source number in each mode are estimated by density peak clustering (DPC). Finally, the density-based spatial clustering of applications with the noise (DBSCAN) algorithm is employed to obtain the final source counting results from the DOAs of all modes. Experiments showed that the multimodal fusion method could significantly improve the ability of a single AVS to distinguish multiple sources when compared to methods without multimodal fusion.

Underwater Source Counting with Local-Confidence-Level-Enhanced Density Clustering

“…Therefore, the performance of the density clustering algorithm is enhanced. An analysis of lake trial data is conducted by comparing the proposed local-confidence-level-enhanced density clustering method with the methods in [10] and [18]. The results confirm the availability of the proposed local-confidence-level-enhanced density clustering method in improving source counting performance.…”

“…As a result, the source number estimation result in the period before 44 s was used to calculate the source counting accuracy. A comparison was conducted between the proposed local-confidence-level-enhanced density clustering method, the multimodal-fusion-based method in [10], and the basic density-clustering-based method in [18]. As shown in Figure 11, the basic densityclustering-based method obtained 48.82% accuracy of target number estimation, the multimodal-fusion-based method obtained an accuracy of 51.15% with four modes employed, A comparison was conducted between the proposed local-confidence-level-enhanced density clustering method, the multimodal-fusion-based method in [10], and the basic density-clustering-based method in [18].…”

“…Acoustic vector sensors (AVSs) can co-pointedly and synchronously measure sound pressure and particle velocity vectors [1][2][3][4]. AVSs are widely used in underwater target detection, direction finding, tracking, and communication [5][6][7] due to their abilities of spatial filtering [8], DOA estimating [9], and resisting the interference of isotropic noise [10].…”

“…For this reason, a multi-source detection based on multimodal fusion was developed in [10]. The output of the AVS was decomposed into multiple modes by intrinsic time-scale decomposition (ITD), so that the source number in each mode decreased.…”

A Data-Mining Interpretation Method of Pavement Dynamic Response Signal by Combining DBSCAN and Findpeaks Function

“…While the types of data targeted are various since they are collected by different road sensors [16][17][18], the main function of the aforementioned methods is to automatically classify data, which is currently the most common approach for abnormal data diagnosis. The DBSCAN clustering algorithm is a representative density-based clustering method that utilizes a set of neighborhood parameters (Eps and MinPts) to define the set's sample density [19,20]. This will lead to the division of sample data into distinct clusters, the identification of noise as well as the detection of outliers.…”