Department of Mechatronics Engineering, Bahcesehir University, Istanbul, Turkey.
J Acoust Soc Am. 2011 Apr;129(4):2059-67. doi: 10.1121/1.3557031.
Approximately a quarter of all West Indian manatee (Trichechus manatus latirostris) mortalities are attributed to collisions with watercraft. A boater warning system based on the passive acoustic detection of manatee vocalizations is one possible solution to reduce manatee-watercraft collisions. The success of such a warning system depends on effective enhancement of the vocalization signals in the presence of high levels of background noise, in particular, noise emitted from watercraft. Recent research has indicated that wavelet domain pre-processing of the noisy vocalizations is capable of significantly improving the detection ranges of passive acoustic vocalization detectors. In this paper, an adaptive denoising procedure, implemented on the wavelet packet transform coefficients obtained from the noisy vocalization signals, is investigated. The proposed denoising algorithm is shown to improve the manatee detection ranges by a factor ranging from two (minimum) to sixteen (maximum) compared to high-pass filtering alone, when evaluated using real manatee vocalization and background noise signals of varying signal-to-noise ratios (SNR). Furthermore, the proposed method is also shown to outperform a previously suggested feedback adaptive line enhancer (FALE) filter on average 3.4 dB in terms of noise suppression and 0.6 dB in terms of waveform preservation.
大约四分之一的西印度海牛（Trichechus manatus latirostris）死亡是由于与船只碰撞造成的。一种基于对海牛发声被动声学检测的船只警示系统是减少海牛-船只碰撞的一种可能解决方案。该警示系统的成功与否取决于在存在高水平背景噪声的情况下，特别是在存在船只噪声的情况下，对发声信号进行有效增强。最近的研究表明，对噪声发声进行小波域预处理能够显著提高被动声学发声探测器的检测范围。本文研究了一种自适应去噪程序，该程序基于从噪声发声信号中获得的小波包变换系数实现。与单独使用高通滤波相比，当使用具有不同信噪比（SNR）的真实海牛发声和背景噪声信号进行评估时，所提出的去噪算法可将海牛检测范围提高 2 倍（最小）至 16 倍（最大）。此外，所提出的方法在噪声抑制方面平均比先前提出的反馈自适应线增强器（FALE）滤波器高 3.4dB，在波形保持方面高 0.6dB。
