小波域子空间滤波法研究  

Study of Wavelet-Domain Subspace Filtering Approach

作　　者：王金芳[1] 杨宝俊[2] 李月[1] 王金宝[3] 

机构地区：[1]吉林大学通信工程学院,吉林长春130012 [2]吉林大学地球探测与技术学院,吉林长春130026 [3]东北师范大学地理系,吉林长春130024

出　　处：《大连交通大学学报》2008年第2期36-41,共6页Journal of Dalian Jiaotong University

基　　金：国家自然科学基金资助项目(40574051)

摘　　要：将时域子空间法和小波变换结合在一起,推出小波域子空间滤波算法.具体地讲,把整体观测信号小波系数子空间划分为信号加噪声小波系数子空间和噪声小波系数子空间,根据尺度内时间估计准则,由信号加噪声小波系数子空间重构信号小波系数,从而获得估计信号.实验表明,当用于恢复高斯噪声中Ricker子波时,小波域子空间滤波法相对于传统软、硬阈值法具有明确优势.Subspace filtering scheme in the wavelet domain is proposed by combining time-domain subspace method with wavelet transform. Concretely, the total subspace of the observation wavelet coefficient is partitioned into the subspace of signal-plus-noise wavelet coefficient and that of noise wavelet coefficient and the signal wavelet coefficients are reconstructed by the subspace of signal-plus-noise wavelet coefficient based on the time constraint estimate within a scale （TCS） to obtain the estimated signal. The experiment results show that the algorithm possesses the definite advantage compared with the traditional soft or hard thresholding when recovering Ricker signal corrupts by Gaussian noise.

关 键 词：噪声消除 子空间法 尺度内时间估计 

分 类 号：TN713[电子电信—电路与系统]