小波域中用多尺度滤波器提取探地雷达信号  

Extraction of Ground Penetrating Radar Signals by Multi-scale Wiener Filter in Wavelet Domain

作　　者：詹应林[1] 昌彦君[2] 邱宁[2] 

机构地区：[1]中国冶金地质总局中南地勘院,湖北武汉430074 [2]中国地质大学地球物理与空间信息学院,湖北武汉430074

出　　处：《资源环境与工程》2008年第1期85-88,共4页Resources Environment & Engineering

摘　　要：针对实际中观测到的探地雷达信号时常会被噪声污染的情况,根据小波分析和分形理论在多尺度分析和自相似本质上的一致性,将多尺度维纳滤波器理论引入探地雷达信号处理中,提出基于双正交小波变换与维纳滤波的多尺度滤波算法,提取探地雷达1/f类分形有用信号。首先,利用双正交小波变换将带有加性白噪的探地雷达信号分解成多尺度的子带信号,通过小波变换对1/f类分形信号的白化作用,消除了1/f类分形信号的非平稳性、自相似性和长程相关性。然后,在多尺度小波域中利用维纳滤波,实现噪声和有用信号的分离,估计出各子带中探地雷达1/f类分形有用信号。最后,利用双正交小波的精确重构性,恢复加性白噪背景下的探地雷达1/f类分形信号。实验表明,该滤波器能有效提取探地雷达有用信号,显著地提高信噪比。In terms of the identical properties of the wavelet analysis and fractal theory in multi - scale analysis and self similarity, the authors introduce the theory of multi - scale wiener filter into the treatment of ground penetrating radar signals, extract available 1/f fraetal signal from ground penetrating radar based on biorthogonal wavelet transformation and calculation of multi -scale wiener filter. First, the signal with white noise is transformed into multi -scale subsystems through biorthogonal wavelet transformation. Unsteadiness, self- similarity, long- range eorrelation of 1/f fractal signal can be eliminated through whitening activity. Then, the Wiener filter can be applied to estimate the available signal from each sub - band. Finally, the 1/f fractional signal interfered by white noise is resumed through reconstruetion of biorthogonal wavelet. The simulation experiment shows the wiener filter can extract signal from ground penetrating radar effectively, and improve signal to noise ratio remarkably.

关 键 词：小波变换 1/f类分形信号 维纳滤波器 探地雷达 

分 类 号：P631.443[天文地球—地质矿产勘探]