摘要
首先介绍了小波阈值消噪的基本原理,认为小波阈值消噪的性能受三方面因素的影响:(1)小波母函数的选择;(2)阈值方法的选择;(3)阈值函数的选择。为了寻找小波母函数、阈值估计和阈值函数的最优组合,文中以实测海洋磁力测量数据为背景,采用小波阈值法进行消噪,把得到的磁异常值作为"真值"。仿真了波浪噪声和随机噪声并加入到"真值"磁异常中,作为噪声磁异常。在此基础上采用两种小波族系(Symlets,Daubechies)、四种阈值选取方法和软、硬阈值函数对噪声磁异常进行了基于小波变换的消噪计算,客观分析了不同组合方法的消噪精度和可靠性。实际计算结果表明:对磁异常突变点较多且分布杂乱的测线进行消噪,Sym6、软阈值函数、混合阈值或基于Stein无偏似然估计消噪效果最好;对磁异常突变点少且大部分光滑的测线进行消噪,db7、基于Stein无偏似然估计和软阈值函数的消噪效果最好。
A magnetic anomaly with wave noise and random noise is simulated based on practical data of marine magnetic survey to evaluate the performance of wavelet threshold denosing via different combinations of mother wavelet functions and thresholds. Mean square error is defined to assess the accuracy of denoising results. Two families of mother wavelets （Symlets, Daubechines）, four threshold selection rules and two threshold functions are tested in a series of calculations to estimate the functions of these wavelet and threshold parameters. By calculation, it is shown that in the case of lots of anomaly and pockety survey line, the best denoising performance is reached via the combinations of sym6, soft threshold function and rigrsure threshold selection; in the case of little anomaly and smoothness of survey line, the best denoising performance is reached via the combinations of db7, soft threshold function and rigrsure threshold selection.
出处
《海洋技术》
2008年第3期66-69,79,共5页
Ocean Technology
基金
国家自然科学基金资助项目(40671161)
国家专项基金
关键词
海洋磁力测量
小波变换
阈值消噪
marine magnetic survey
wavelet transform
threshold denoising
