吊放声纳信号提升小波阈值去噪分析

Lifting Wavelet Threshold De-Noising for Dipping Sonar Signals

为了去除干扰噪声及保留信号特征,本文采用提升小波阈值法对信号进行去噪处理。先对信号进行提升小波分解,以获取相应尺度的尺度系数和小波系数;采用相应的阈值函数对小波系数进行量化,将量化后的小波系数和尺度系数按照提升小波逆变换方法重建信号,获取去噪信号。仿真结果表明,该方法去噪速度快、占用空间小,去噪信号信噪比均在10dB左右;均方根误差和峰值误差分别控制在0.1和0.3以下,波形与功率谱趋于平滑,信号的峰值点得以保留。该去噪方法在保留信号基本特征的同时,有效抑制了干扰噪声。

In order to reduce noises and keep signal feature, lifting wavelet threshold de-noising was ad- vanced. Firstly, the signal containing noises was decomposed with lifting wavelet and the corresponding scale and wavelet coefficients were gained. The improved threshold function was used to quantize the wavelet coefficients. The quantized wavelet coefficients and scale coefficients were taken to reconstruct signal with inverse lifting wavelet and de-noised signal was got. The method proposed here has a great computational speed and a lower memory requirement. The time domain wave, power and Demon spectrums of the de-noised signals were compared with the original ones. It indicates that the noises are reduced while the steady state character and peak value features are reserved. This de-noising method suppresses noise effectively and retains basic characteristics.