F-X域奇异值分解预测滤波法随机噪声衰减

1998年 37卷 第No. 3期
阅读:88
查看详情
Random noise attenuation using predictive filtering in F-X domain by singular value decomposition
1. 大庆石油学院, 安达 151400;2. 大庆勘探开发研究院, 大庆 163712
Daqing Petroleum Institute, Anda 151400
信号分析中的预测技术在地震勘探中得到了广泛的应用, 一维预测滤波能消除鸣震等具有周期性的干扰波, 但对随机干扰则无能为力。针对这一困难, 最近几年出现了F-X域二维预测滤波。该方法是将时间信号变到频率域, 对每一频率信号横向上进行预测, 由于有效波横向上是相对连续的, 而随机干扰在横向上是不相关的, 因而能有效地消除随机干扰。
The predictive technique in signal ana1ysis has found extensive application in seismicexploration-The l-D predictive filtering can remove Periodic disturbing wave, but is powerless toremove random disturbance. In view of the difficulty, the 2-D predictive filtering in F-X domainhas been proposed in recent years. In the method, we trans form the signal from the time domaininto the frequency domain, and predict each frequency signal in the lateral direction. As theeffective wave is relatively continuous in the lateral direction and the random disturbance isuncorre1ated in the 1ateral direction, therefore we can remove the random disturbance effectively.
F-X域; 二维预测; 滤波因子; 随机干扰; 奇异值分解;
F-X domain; 2-D prediction; filtering factor; random disturbance; singular value decomposition;